Warning! The site is under development. There might be some functions that doesn't work properly.
This is a mascot for a women's clothes online shop and its mobile application. According to the Design Specifications, we have to create a 3D character with a changeable by customers appearance to make avatars for themselves. The textures should be detailed enough for the middle and long shots. The appearance change should be implemented by replacing one of the rendered mascot passes.
Initially, the client didn't require a video, but we've done some project reproduction for our portfolio and we've loaded the mascot into Unreal Engine.
The current price represents more complexity of the 3D animation than this example because of the constantly increasing software and hardware advances.
The character prototype is a stylized 20-25 years old woman, average shape without any deviation from athletic or overweight. The other shapes were shifted between 61-69 inches in height and from XXS to XXL clothes sizes with different hair cuts, hair, skin and eye color. She wears jeans, a T-shirt and gumshoes.
At that moment, we didn't know where the character would be used, so it would be in a T-pose and rigged enough to change it quickly, thus we decrease further edit time. There is a bottom line of 576 appearance variations.
Using the above mentioned data we created a 3D mascot concept. The stage helps process most of the edits, and take into account the client's request for the lowest possible price. With concepts, we get the expected result without blind-box actions, and it helps the modeler to go along with proportions. Also, we will check the appearance’s variation.
Unlike straightforward 3D character sculpting from a primitive, we used a blocking method. If one uses simple geometry shapes, it is possible to change the body faster with more freedom. The main idea of this method is adding details step by step, thus the client can see the result from the first day. The maximum details level depends on Design Specifications and shouldn’t change during the project because it means a lot of unnecessary edits.
When the prototype shape was ready, we merged primitives and smooth junctions to get the finished look.
After receiving the positive review for the first stage, we created clothes, shoes, and simplified hair to check all the model’s parts together. Also, we added wrinkles on the clothes that are better to do now than during the texture stage.
The main difference between the animated models from the static ones is the topology’s ability to change pose correctly, when the volume stays the same. Changing the 3D model to use bones and blend shapes is required by the Design Specifications. The better the modeler knows topology, the less extra work we need to do in further stages, especially in rigging. It’s important to consider how the mascot will move and what we need to do using textures or what just the topology. The higher the topology dense, the slower the animation process. At that stage, hair didn’t matter, it would be replaced with generated one.
After making a simple rig to pose the 3D character, we made three generated haircuts. In the first project's version, we used X-gen and faced overkill crashes trying to change hair's length and curves' position, so we adapted, abroad of time tracker, the workflow to accomplish the project.
Last few years, we have created a flexible workflow that allows us to make and change hair fast and accurately, disregarding how the character moves, including force fields and clothes simulation. We have made the second project's version specially for this article by using Unreal Engine to render a video example. During this process shaders were made for skin, hair, clothes, and eyes, which with proper hi-res texture increases quality up to photorealistic.
The project's outcome was adapted for the mobile platform however, we needed to take into account further resolution updates on smartphone screens and make a small margin. Downscaling the final image was mighty cheaper than re-rendering. We used UDIM instead of making shaders separately, so we made the main shader and then detailed it by adding wrinkles, seams, and scuffs.
It was one of the most enjoyable projects, from the very first conversation with the client to the result. During the unique 3D project, one can find a lot of little troubles which are needed to overcome to get a significant result, and this one wasn’t an exception. There were the technical limitations of X-gen, and we learned how fragile and unyielding the plugin was. We were getting a Maya crash after almost every step aside from the simple hair setup. At moments like these one, we just turn off the time tracker and try to derive a smooth way. After a while on the project, we’ve developed our fully flexible and stable pipeline to create 3D characters for animation and still render using the most modern ways and applications.