Perceptual plausibility of exaggerated realistic motion
Squashing and stretching our way across the uncanny valley.
Abstract
The informal heuristic practices of the fine arts have much to offer to our understanding of the appearance of phenomenological reality. One interesting example is the use of exaggeration to enhance the illusion of liveliness in both living and nonliving subjects. This further eases the uncomfortable sense that the motion is somehow uncanny — especially with inanimate objects. We performed a series of experiments to test the effects of exaggeration on the phenomenological perception of simple animated objects — bouncing balls. A physically plausible model of a bouncing ball was augmented with a frequently used form of exaggeration known as squash and stretch. Observers were shown a series of animated balls, depicted using systematic parameterizations of the exaggeration model, and asked to rate their plausibility. A range of rendering styles provided varying levels of information as to the type of ball. In all cases, balls with small amounts of exaggeration were seen as plausible as those without any exaggeration (e.g., with veridical motion). Furthermore, when the type of ball was not specified, observers tolerated a large amount of exaggeration before judging them as implausible. When the type of ball was indicated, observers narrowed the range of acceptable exaggeration somewhat but still tolerated exaggeration well beyond that which would be physically possible. We contend that, in this case, exaggeration acts to bridge the so-called uncanny valley for artificial depictions of physical reality.
Schmidt, F., Noejovich, L., Chakalos, G., & Phillips, F. (2024). Perceptual plausibility of exaggerated realistic motion. Cognition, 251, 105880. https://doi.org/10.1016/j.cognition.2024.105880
Turdpaint
In my computer graphics youth, everyone wrote two programs – a 2D paint program and a 3D renderer.
Such was the curriculum of the time. Ed Tripp taught CIS 781-2-3, a three quarter sequence that took you from the Bresenham line algorithm1 2 all the way to a scanline renderer. I wrote 99% of my code for that class on an original Macintosh 128K, in Megamax C and Macintosh Pascal.
For the in-person final exam for 783, Ed came in, wrote “Final: Write a ray tracer in JCL” on the board before we got to class. Having written a little bit of JCL in my life (we had a IBM 360 emulator that ran on the Amdahl, I learned assembly language on that) and I knew of its unreal power I was slightly panicked.
JCL or “Job Control Language” was the stuff you put before your actual code to tell the computer what to load, where to load it from, where to get input and output, what to do with the code you gave it, etc. The only thing I remember from back then was forward slashes, and that DD
was the data deck, so you’d say something like –
//SYSIN DD BLEH
to get it to load your data deck. It was a crazy but necessary part of getting things to run on mainframes.3
Luckily, it was a joke. But, you know, for a minute, it wasn’t.
When I finally got access to a color Macintosh (a Macintosh IIsi if I recall correctly, it was an engineering sample I got @ Pixar) I modified my paint program to work in color. The new version had a bug that caused it to paint in splats4, little blobs, and was christened “turdpaint” by my wife.
I gave a talk at some point on user interfaces for computer graphics (I think it was for the Human Factors Society at Ohio State) probably around 1992 or 1993. I found the above image in the slide deck from that talk. The paint brush cursor was the best thing about it, now in color, and the icon was my second favorite thing –
lest we forget that I went to a Bauhaus-Fascist5 architecture school.
Anyway – I found the slide deck, in which there are some really great images, but thought about turdpaint and its color revival, and it brought me great joy.
- I was on a SIGGRAPH bus, on my way to a social event at the ranch they used for the TV show Dallas.The gentleman next to me (who was wearing a very nice cowboy hat) introduced himself as Jack Bresenham. I said “Flip Phillips, I use your algorithm.” We had a great talk for the bus ride. ↩︎
- https://resources.wolframcloud.com/FunctionRepository/resources/BresenhamPoints/ ↩︎
- Fred Brooks, supervisor of the OS/360 project, for which JCL was created, called it “the worst computer programming language ever devised by anybody, anywhere”. (see Wikipedia) ↩︎
- A side note, I write a 3D particle renderer at Pixar that used what are now called Gaussian splats, back in 1989. I think Bill Reeves wrote one even earlier that inspired mine. ↩︎
- That characterization thanks to Tony Lupidi, a painter a great friend at CGRG who was as non-Bauhaus as I was Gropius. ↩︎
Papa’s Draft Card
Found my Grandfather’s draft card in a bunch of my Dad’s stuff.
About a year before my Grandfather passed away, I visited him and he pulled his draft card out of his wallet. He handed it to me and said “I keep trying to get the shuttle bus driver to drop me off at the post office so I can burn this on the steps. He won’t ever stop.”
That is some serious joke long-game, this was in 1992 or so.
I’m going through a lock box of my Dad’s stuff (no gold bullion… dang) and found it, laminated even. Dad must have done that (he loved laminating things) after Papa died. I like to think that he kept it laminated so he could re-tell the joke.
Oh the things you’ll find…
I’m cleaning up a bunch of archived files.
While I was in grad school, after Pixar, I did a little bit of consulting to make ends meet. One of the more fun things that I did was write a bunch of audio software for a David Bowie interactive CD. (aka CDi)
I got to hang out with Brian Eno, visited him on Guy Fawkes Night while I happened to be in the UK (His neighbors really liked him, he’s a nice guy.)
Somehow, I seem to have come upon the code for one of the things I wrote –
I have a DAT with Black Tie / White Noise tracks. You could remix Bowie in real time. Man that was fun.
Amazingly, you can still buy a used version on Amazon – https://amzn.to/3x441Nn
Harsh review :)
MacRenderMan
Some historical floppy disk fun from old Pixar days.
When Pixar decided to sell RenderMan to the public we had to come up with a good name (It was REYES, Renders Everything You’ve Ever Saw / Seen). You could already buy a version that a ran on the Pixar Image Computer CHAP (Channel Processor) called CHAPREYES. I don’t know if we ever sold any, but we did render Red’s Dream on it, in part.
The internal code name was RenderMan. Jeff Mock and JL designed some ‘Superman Logo’ shirts and the whole shebang. I wish I still had mine. It went to the car-wash bucket a long time ago, along with my original Surf Ohio shirt.
We had a contest to name the new product but RenderMan won. So that’s what you’ve got. It ran pretty slow on the Macintosh of the time, but it ran. There was no UI (You had to buy ShowPlace which was a sort of Adobe Dimension of its time) but I wrote an interface to Mathematica so I could render things for my dissertation work.
When Pat Hanrahan sent out the original RIP specification this wonderful note came back –
Trivia Note! It was originally RenderMan Interface Protocol, and the files were .rip
files, but Adobe was starting to do that PostScript thing for Raster Image Processor files, so we changed to RenderMan Interface Bytestream or .rib
.
Also, it got a nice IEEE award 30+ years later, so always play the long game, folks.
Above, the original RenderMan ‘bouncing r’ logo as a pin. Mint in package, well, mouse-eaten package.
Chernobyl Personal Geiger Counter
Issued to residents of Chernobyl and surrounding areas after the Chernobyl Disaster.
Last I played with it, it still worked. It takes some pretty weird batteries and I think a coil has cracked inside, so I need to do a little renovation work on it.
I got it from a gemologist friend. And I am sticking to that story.
If you’re interested in some interior photos, let me know. I’ll grab some when I take it apart to fix the coil.
On Talks and Slides
Some history of style.
Once upon a time, Ed Tufte called me a “sanctimonious assohole”. I suggested that he up the dosage of his medications and stopped recommending his books to my students and colleagues1. But — The Cognitive Style of PowerPoint remains a pretty good essay (sometimes excessive) on some of its ills.
I grew up in a Kodak Slide Carousel era of presentations. Every ‘big’ talk I went to as an undergrad was formatted that way – Ektachrome 35mm slides. Usually of images — text was actually difficult to put on them, you needed a copy stand or a slide printing rig2. Now you just type words and, boom, you’re done. Add some clip art of puppies or maybe some insane typographical effects and you get the sort of talk everyone loves to hate.
At Pixar, we had a text generating system for the Image Computer. It was based on the fonts that Don Knuth had put together (Metafont) and a version of roff/nroff/troff
by Tom Porter, if I’m remembering right. Here’s one from a talk I gave in London in 1989 —
Why that title? We sat around and decided to call ourselves, the animation group, “Studio Pixar” to identify as a different group than was making hardware and software. It sounds a little silly in retrospect, but hey, we weren’t in branding, just goofiness. That image rendered on the Pixar Image Computer and was then transferred to Ektachrome or 5247 using the Pixar Laser Scanner3, processed and mounted.
In those early days, a subset of Eben, Pete, Andrew and I used to go to places (usually colleges, usually Berkeley and Stanford) to show our films and talk about how we did things. This was all new stuff then, people wanted to know how we did it. Before one adventure, Pete decided to make some title and word slides that were more visually interesting to go with our more visually interesting content. I drew a few too, but Pete’s were the majority of the talk —
I found them a while back. I love them so much —
This was a nice vacation from these —
Anyway, someone was asking why I use so few words on my talk slides. I use a lot of (simplified) graphs and a lot of photos. I personally think it helps with the storytelling — to not have people reading lists of words, tables of numbers, &c. I loved the beamer
LaTeX class back in my dissertation days — my slides were literally part of the document I created. Knuth’s Literate Programming, from which the Mathematica Notebook4 took inspiration, was also an important of my presentation upbringing. One document, code, paper and talk slides, all in one. It all comes back to Knuth somehow, which I personally think is a good thing. A very thoughtful guy who facilitated a lot of what we use today.
Finally — if you’re looking for a tool to create more thoughtful presentations, have a look at iA Presenter. Markdown driven slide presentation — like Markdown, TeX, and roff, thinking about your *content* rather than which awesome transition and bullet point animation you’re going to use. Highly recommended process.
- At a particular institution I was associated with, I put the kibosh on a plan to retain him as a consultant on visual literacy. I showed the committee the correspondence between him and me and they decided that they didn’t want such a fragile personality involved in the project. It’s the little sanctimonious revenge that warms the cockles of the heart in the wintertime. There was also possibly a bit of pot-kettle going on here I think. Regardless, I hope he’s feeling better these days. ↩︎
- Does anyone else remember the SIGGRAPH where, somehow, Larry Yeager (I think?) didn’t have his slides and made a set of hand-written bullet points, photographed them in his hotel and had them rush processed before his talk? Maybe I’m hallucinating that but I don’t think so. ↩︎
- Another interesting piece of IO hardware, built by David DiFrancisco. ↩︎
- Don’t get me started on some takes on the ‘innovation’ that are Python notebooks. I’m super happy they exist, but they’re not the new crazy thing everyone thinks. ↩︎
The PICS 2000
A medical imaging workstation – the beginning of volume imaging
When I got my interview at Pixar, I was at Ohio State working on volumetric rendering of CT and MR images. There are a few callbacks to that on here, OSU MRI Flashback – 1984 for example. We were at the leading edge of that stuff back then. Pixar was looking for commercial sales of its Pixar Image Computer and volume imaging was a perfect use for a gigantic frame buffer + computer like the Pixar. This used the Pixar II, the second generation, designed to be a little more affordable. (It also implemented what was to be the future of high-res TV standards and other little fun tidbits.) It used a consumer-grade VHS recorder and a modified version of Steve Wozniak’s CL-9 CORE universal remote control. I learned how to write 6502 code for that thing, and wrote a nice little controller library. Bruce Young of our hardware group and I made a few trips down to CL-9 in Los Gatos, learned about this revolutionary-at-the-time device to control multiple devices with a single remote control!
The PICS had a SUN, Pixar II, and a 9-track tape drive + two monitors (one for the Pixar, one for the UI). I was tasked to design the UI. Back then, there wasn’t really much well-established UI/UX practice, so I mainly looked at the Macintosh and MacOS for inspiration. We used a spatial layout, moving ‘tasks’ like reading tape, processing images, outputting renders, from left to right in an inverted-U shape.
We built all sorts of neat tools for segmenting the images (my introduction to Bayesian statistics via Pat Hanrahan) selecting ROIs, etc. It was a really cool device and _way_ ahead of its time alas. I made two trips to RSNA (more good stories there about taxis, young Stephen Colbert @ Scuzi restaurant, a dead man’s Model Railroader, and more!) to help try to get radiologists to think they needed this beast, and, indeed I think we sold a total of six. The industry wasn’t ready for volume rendering. Now, when you pick up a CD from your radiology center, it’s got a volume renderer baked-into the viewer already on the disk.
There was not much in the way of agreed-upon medical imaging networking (for history fans, PACS was just being discussed at various imaging meetings, and wasn’t near being settled work) so you’d take the scan of the patient on 9-track tape and load it into our machine, read, and decode it. I had been doing this at OSU with the GE hardware we had there, and even for my Father-in-law’s pharmacy business, to read 9-track tapes which were the currency of exchange for medical (and, of course, other) data at the time.
I was digging through my archives, moving from house to house, and found the original literature / flier for it. And now, it is yours to enjoy too –
If you ever got to see one of these, or the CL-9 (Cloud-9 alias) CORE or anything else fun like this, let me know.
Highland Aire
I just finished a row. It’s about 99% humidity this morning, which, to credit Chris Wedge, reminds me of having just rowed inside someone’s mouth.
Chris and I used to ride our bikes together when we were at CGRG. That was back in my racing days, I was decidedly a mediocre racer1 but I had a lot of fun. That was right when Greg Lemond became the great hope, 7–11 flew the US flag, I met Norman Alvis from that team, Motorola, Postal, and Saturn, who, at 60, is about to try to regain the hour record, which he held for years.
Chris referred to Pat Metheny and Lyle Mays as “Dentist office jazz”. I suppose I respect his opinion and can understand. After all, he lived next door to Kathleen Brenan for a while.
Highland Aire from Lyle Mays just came on random. It reminded me2 of one day back in the late 80’s, I had to go to San Francisco for some reason and, on my way back up to Fairfax, there was a traffic jam on 101 that started at the bridge3. Since I used to ride over Mt. Tam I decided to just take the long, scenic way home.
Right when I hit the Panoramic Highway4 that tune came on. It was, in context, perfect. Pretty much a theme song for that stretch of road.
Which further reminded me — another old friend I miss from those days, Rich McKay, was working on a crew on the Panoramic as AC on a (newly introduced!) Miata commercial. Wilson Burrows, also CGRG / Marin alumni, once told me that, if you watch car commercials, every 3rd or 4th one is shot on that road. I remember having to ride through a shoot once, to the chagrin of the PAs because it was closed for the shoot.
Sorry man, if I was in a car, I’d turn around. But I’m on a bike and I’ve got places to be.
Anyway, the shot was of a wedding party celebrating. There was to be the release of birds and the Miata was to drive through the flock, forced perspective in depth, to give the birds safe clearance, etc. On the first take, apparently one of the birds wasn’t feeling very well and, when released, just sort of flew ‘down and toward’ the car as opposed to ‘up and away’. A greatly unintended collision occurred between the ill bird and the car.
The crew and actors were stunned and depressed. The grill was damaged. Shooting was called off for the day.
And that’s what I think about when I hear Highland Aire.
- I was never going to get out of Cat 3. Ever. ↩︎
- It also reminded me of when I had just discovered Bill Frisell w/ Jan Garbarek, and Stephen Spencer, also CGRG, who I told about. And how we bonded over it ever since. ↩︎
- I realize that is common now, and it happened frequently back then, but not always, like now. ↩︎
- Panoramic, but hardly a highway. ↩︎