Digi-Double CG Dog
This project was in memory of my dog Maggie who passed away last Christmas. The purpose was to preserve her as a CG digital replica. It was also my final project whilst studying Visual Effects at Norwich University of the Arts and my first attempt at the entire CG Character Pipeline.
Hello! Welcome to my entry to the 2022 Rookie Awards for Visual Effects. My project was for my final Year studying BA Hons Visual Effects (VFX) at Norwich University of the Arts, creating a CG digital replica of my dog who sadly passed away Christmas day 2021. I have created this in loving memory of Maggie.
I have focused on photo realism and relying heavily on reference imagery for this project aiming to recreate her as accurately as possible. I have created a detailed breakdown of the entire project which has taken me about 5 months from start to finish. This was my first attempt using ZBrush to model a character and XGEN to create the groom. By carrying out the entire character production pipeline I have solidified the direction of my career path towards creature modelling. I hope you enjoy seeing the breakdown I have created. The final results are compiled together in my 2022 Showreel at the end of this page.
Full Groom Render with Wire-frame Overlay
Close Up of Eyes
Half Groom Half Sculpt Photoshop Edit
Perspective view Renders
Turntable of Groom on Static dog
Below shows a breakdown of the key stages of the pipeline and what tools and techniques I have used to achieve the final results.
I began the modelling process in ZBrush, blocking out the forms from a Basic Clay sphere. Previous to this project I had only practiced basic anatomy in the software, so this was my first attempt creating a photo realistic character from reference.
Here are the various orthographic views of the sculpt once I was ready to take it into Maya to begin the next stage of the pipeline - re-topologizing the mesh.
The sculpt in ZBrush was exported as an OBJ as separate subtools, starting with the main body of the dog. This was imported into Maya to begin the retopology process which was essential to creating a character that can be rigged and animated effectively.
Here are the final results of the Retopologised mesh. This is my LOW POLY version of the model with as few quads as possible. I am really pleased with how it has turned out, particularly the face. Although it took a lot longer than anticipated, I now have a much deeper understanding of how to go about the process next time.
Creating the UV's
After retopology was complete, I was able to create and unfold the UV shells that would enable me to texture the model. I wanted this to be a hero asset so the map sizes needed to be enough for 2-4k resolution, particularly on the face. I grouped each material into separate UV tiles, working in the UDIM workflow. This process turned out to be a lot quicker than expected and I didn’t face any issues.
In order to combine the detailed original sculpt from ZBrush with the new Retopologised version from Maya, I used re-projection to apply the details to the LOW POLY mesh.
Here are some rendered views of the Low Poly wireframe of the dog model in Maya. This was created using an aiWireframe Standard surface and ensuring the smooth preview was not enables in the viewport. This is the version that I sent to be rigged and animated as it has clean topology for them to add edge loops were necessary and keeps the scene light and workflow more efficient.
Sculpting the Tertiary Details
Once the reprojection process was complete, I was then able to sculpt high frequency details such as veins and wrinkles onto the dog model in ZBrush. I subdivided the model several times to ensure I could achieve a high level of detail from reference.
In order to apply the high frequency details created in ZBrush, I would need to drive them with displacement maps. This will keep the scene lighter to accommodate for the animation and groom by keeping the LOW POLY version of the dog model.
Rigging and Animation
The rigging and animation of this project were outsourced using an online service by two separate artists. I provided them with the LOW POLY version of the dog model in the static T-pose and gave a brief overview of the purpose and requirements of the rig.
I used Substance Painter to texture each subtools as this allowed me to work simultaneously as the rest of the pipeline and update the maps as further iterations were made.
As this was my first time creating realistic hair using XGEN, I began by doing a lot of research and analysis into what makes hair look photo-real. By understanding the properties I was able to create a more realistic groom. I did various studies of my reference imagery to visualise where I would place the guides on the model.
I created all the guides in the XGEN Editor and then mirrored them across the line of symmetry using the object topology. From this I then had take the guide information to be able to apply the density and colour maps.
I wanted to add even more detail to the dog based on the reference imagery so I began to create new interactive groom descriptions for thinner white whiskers and eyelashes. Although these wont be that noticeable on the overall body shots, it will make the close face shots a lot more realistic.
Animated dog with Groom
This is a version of the groom I created to apply to the animated mesh, it still requires some tweaking but shows an example of the pipeline being completed.
This is my graduate show reel at the end of my 3-year course in BA Hons Visual Effects at Norwich University of the Arts, showcasing my main projects from the last 2 years. The first minute of the showreel is an excellent summary of this project and presents it in a professional manner with further breakdowns.
Thank you for taking the time to look through my work!
Check out my website for more: https://ameliasturdycreations.com/