People who are kind of elevated because of celebrity, looking at them as a fellow human being and letting other people into that is infinitely more interesting than just simply showing another celebrity view.
This element of our animatic features line drawings of celebrities that Colin Davidson has painted. After my earlier tests into the method, I began to work on creating models that matched the celebrities they were of. The first us was Glen Hansard.
What’s important to note here is that this model does not need to be perfectly like Glen Hansard, but it should have a certain likeness. The reason for this is because our line drawings are representative of celebrities, but not totally like them. Like in Colin’s quote, we want to create a view that’s different and more interesting than every other celebrity view. Our line drawings will be rough and sketchy similar to Colin’s early drawings, with transparent cut out alpha channels. For this reason, the models do not need to have fine definition and total perfection, as the mesh will not be seen, only the line drawing over the mesh.
I began by using an image to reference the model, and soon realised that there were better ways to do this. I then set up image planes of Glen Hansard in Maya after watching a tutorial. I decided to model him in three parts; the head, the hair and the beard. This is because I wanted to give the hair volume, as well as a different type of texture style than that of the skin.
I wasn’t making great progress with my model so I turned to another couple of tutorials to guide me. I also took a look at this anatomy guide that Mike Bass had shared, which guided me to study the structure of the face more than I had before. This aided my modelling especially in refining the cheek and jaw area. It was important to be reminded of best practice in the process, and I was able to move forward faster and with more accuracy than before.
After creating a model I was happy with I moved on to the beard area where I merely added a few lumps and bumps. I did this because I wanted to test it quickly, to understand the level of detail that I would need to create an effective model for the purpose of the piece. There would be no point in spending time on it if it was going to work well at it’s simplest version. To test this, I drew up the first more considered line drawing in Photoshop for Glen to see how it would feel in the Oculus. Creating one drawing for the head, and one for the beard, I added the textures to the FBX models I exported and imported into Unity.
When I tested in the Oculus, the heads felt really impactful, but I immediately knew that I needed to add more to the beard; it felt completely flat against the model. I also tried a different style of beard using much more geometric shapes, to test and see if this would make a difference. While it had an interesting effect, the beard still felt flat. Obviously the hair was not on either at this stage, which too would add to the impact of the models.
After testing, I went back into Maya to build up the beard and hair models. I again referenced a tutorial to understand how to do this well for Unity. As this is an interactive piece, I needed to make sure it was low poly for optimization. In addition to that, the model is totally transparent so I had to make sure that there were not any overlapping faces, as this would make the texture look strange when it’s added.
When I finished the hair and bead tweaks, I went into photoshop to try a few things. My previous test let me see how different lines styles felt in the Oculus, and I wanted to create something that was half way between what I did before. Previously the line drawing style was quite curvy, which doesn’t reflect Colin’s drawings as well as they could. I tried a few tests with a variation on the density of the lines in each iteration.
The faces with less lines were most effective, as the denser drawings felt overwhelming and left the face almost unrecognizable. I could quickly see that the form was very easily lost if not highlighted in the right places on the transparent model. In light of this, I decided to try a variation in the strokes I was using; keeping a solid single line for the face, but a finer more cross hatched look for the hair.
This version was getting closer to the effect that I was hoping to achieve, but I felt that the density of lines, again, were too much.
This iteration was one of my favourites – I took a large number of the lines from the face away and then created a looser crosshatch. What I hope to do in the next development, is to change the alpha script so that the dissolves have a softer final value.