 Good evening! It is a real pleasure to be here tonight talking about some of the artistry and technology behind Toy Story 4. We wanted this film to fit in with the Toy Story world but also feel like something very new. And today, Ling and I are going to talk about two important challenges of Toy Story 4, the rain and the sets. And Render Man was a vital part of making them both feasible and beautiful. Thanks, Bob. So as Dylan mentioned in the intro, one of one of our biggest challenges on Toy Story 4 was creating these two really incredibly large and complex sets full of unique objects. We've got the antique store and the carnival. There's a lot of complexity and it was hard and we did not use Google Sheets to render it. So in addition to that challenge, you know, it's a story about toys. So we got like Giggles McDimple on the bottom left here, who is about three centimeters tall. So we've added this extra layer of complexity, which is the extreme, extreme detail. To drive that point home, I will show you this video. So this shows off kind of what we call the fractal complexity of Toy Story 4. You know, we've got our big set, Mountain County set. We're going to come in through the carnival. We're approaching the antique store. Here's the front desk. We're going to go through. There's Gabby's cabinet. And now we are going to zoom all the way in. There's a specific booth. This is where Woody first meets Gabby Gabby. There is a mason jar here. Inside of the mason jar are a bunch of fully modeled buttons. So this video, I just took tumbling around in USD view. And USD is awesome because we are all working on top of each other at the same time. You know, we're all in different software packages, but we're able to use USD to compile all that stuff into one scene so that we all have the maximum context that we can get. And bonus, because USD is live referencing all this stuff, we're able to selectively upgrade and refine very specific things on demand, which helped us achieve our aforementioned goals. So in addition to all this tech, it turns out a big part of the equation is to work with your coworkers, which I highly recommend. It's both good for your soul, good for making friends, but it's also actually the most efficient way of working, turns out. So, you know, with USD being able to give us all that context that we needed, we then built our pipeline upon everyone being able to see the same renders every day so that we can collectively decide, you know, what's important, what's not important. And so in the end, we're able to get something like this, which is really no small task. So let's talk a little bit about sets, right? So for set dressing, our tech team built a workflow where the set dressing artists could collectively gather all of our props into a warehouse inside a USD view where they could just kind of walk around and browse and shop for antiques, right? It turns out the art department also really liked this, and so, you know, that really helped us with our workflow. From there, the set dressers could bring the props directly into Maya with a single click and then dress from there, which kind of really sped up our workflow. We also used over 1,100 props that were from our digital back lot, which is a bunch of Pixar models from previous movies that have been, you know, compiled into baked USD with baked P-tec shading. So as you can see here, that meant we could fill this huge 8,000-foot space very quickly so that we could start refining the detail and make it look handcrafted. And so, you know, we're able to render all this stuff with our render man lighting, global illumination, refraction, reflection, and get that feedback about how all these pieces are working together really quickly. And it's going to be even better when we fully integrate render man 22 and 23 so that we can even get all that lighting live. So from here, we can kind of get into that fine-tuned detail of dust and cobwebs. So all of our dust is generated from within Houdini. They are output as points into USD data and then rendered as curves in render man. And then on top of that, we put all of this other disgusting stuff like fur and hair and grains and leaves. So, yeah, can make it really gross if we wanted to. For cobwebs, our set extension artist developed a procedure for kind of creating these virtual spiders and then setting them loose in our antique store to live their little virtual lives. So this seamless integration between Houdini and USD and finally render man meant that all of this work was happening concurrently with dressing, shading, and lighting. So that was really great. And then finally, you know, similar to what Rob was saying and similar to set dressing, we tried to find a way so that our shading artists could get very quickly to a point that was good so that they could spend their time refining to make it great. And so we built a set of these global materials that we collaborated with lighting and art department on how they looked so that they would just work out of the gate when modelers would attach them to surfaces within Maya. And so here's an image of a shot from the movie. This was right at the beginning out of the gate, out of the box shading. We had our global materials and some back lot models as well with out of the box lighting. And so everything is already responding in a plausible way. And so here it is in the final movie and you'll see that, you know, set dressing has made a few tweaks. We've picked up the lighting. There's some dust in there now. You know, shading artists have added final colors, graphics have changed, got procedural and painted aging. And then in particular, that middle prop Gabby's cabinet has gotten a lot of hero treatment. And so you'll see here, you know, we put in all this work, but it's all refinement work. It's to make it better, not just to make it look not broken. Right. And so in the end, you know, with everybody working together and all of our tools being able to support that, we're able to render something really beautiful like this, which I don't think we would have been able to do 24 years ago when Toy Story first came out. So thank you. And I'll kick it to Bob. Thanks, Ling. So for Toy Story 4, our effects team was given the task to deliver 100 shots of rain. And here are a few examples from the movie featuring toys, humans and Andy's house. We knew it would be a challenge making the rain feel like an exaggerated threat to a toy, but still physically believable. We split our rain into several important features. Atmosphere, drops, splashes, and surface interaction. We used Houdini engine in Katana to do the initial look development. All the materials you see here, water, rain, concrete leaves, even woody, use Pixar surface, which of course now comes with Render Man. All the materials were carefully calibrated to give realistic and reliable illumination response. And this allowed effects and lighting to iterate closely and quickly together. So first, the atmosphere accounts for the region occupied by the air, falling drops, and interdrop collisions. If we had created rain for the entire world and then started filming, there would just have been too much to handle. So we created rain only in the region where the camera can see. Just like a real film set, we could control the rain in front of the camera, around the characters, and general background all separately. The regions are then coherent based on the shower head located above the camera. As the frustrum slices through the set, you can see the rain start up and die back off. All right, next, the drain drops themselves. The drops needed to feel realistic, but slam into the small characters. So according to relevant literature, rain drops oscillate at about 170 hertz. And with a frame rate of 24 frames per second, each raindrop oscillates about six times. To reduce the computation, the falling rain in Toy Story 4 is instead made of tubes of rain. These simulate some of the streaking while preserving the look of the original phenomenon. The next step is one of the most important for the toy scale. We wanted to feel those classic splash crowns attached to every drop. We created a library of splashes from Houdini's flip solver. We stored these as animated USD clips that would instance at each hit point. This ensured that there was a direct connection between atmosphere, raindrop, splash, and ripple. USD made the path between Houdini, Katana, and RenderBand flexible and fast. Now, while you can't track every drop visually, the overall impact is of the unified system of rain. Each piece is exaggerated to make the scene feel like it's a real struggle at Woody's 15 inches high. All right, now finally when a drop hits, we want it to be believably interacting with the surface. And this meant modeling the drop trails across hard and soft surfaces. For a window, we wanted to capture the interplay of gravity and surface tension. And in this first rivulet, you can see a simple, randomly generated path. Next, you can see the stop and go motion caused by surface tension. The third rivulet here shows a thicker curve that is missing beating while the fourth rivulet now mixes in that beating behavior. This procedure is applied to several thousands of curves, from the windows of the house to the cars parked on the street. And when we convert those curves to geometry, we see their complex behaviors come together. We developed this inside of Houdini, again building the geometry through to Katana and RenderBand via USD. And in these shots, you can see the result. The great part about Toy Story is that we have full-sized merchandise to use as reference. And from these real-world tests, we learned how differently cloth and plastic interact with rain. And we also learned that you should take the batteries out before you do tests like these. To apply the same technique we did for the windows, we had to find a way to create curves on woody. For this, we relied on a point simulation to create a trail of points. And these trails are then used to generate curves. We added geometry and droplets. And finally, you can see the result in context. And sometimes things don't quite go as expected, but it still looks pretty awesome. We also needed the rain to interact with non-solid surfaces, namely the gutter river. We created physical splashes in the water simulation at hit positions. To ensure that the drops didn't create splashes on existing splashes, we tagged each one with an attribute that decayed over time as they fell back into the water. And we wanted that river to look more like a new flood, dirty and muddy with debris. So to get dirty looking water, we put a murky volume inside of the water geometry with a pattern of density and color that flowed with the current. All these elements choreographed by highly skilled artists produced the look of the rain on Toy Story 4. Render Man and USD helped create a flexible pipeline where those artists could collaborate and elevate the scene. We just want to say a quick thank you to the crew of Toy Story 4, Render Man, and of course, everybody at Pixar Animation Studios. Everything you saw tonight comes down to their hard work and passion. So thank you very much.