Sunday, 17 February 2008

Blend Shapes, Colour Experimentation and Paint Weighting

It took all day but I set up the blend shapes. Pocahontas' were much easier as her face is like a cilp on mask. This is similar to how Sally was done in the Nightmare Before Christmas. Cornelius was a bit more difficult since I discovered that some vertices were not merged properly, so I had to go back and re-merge them. The first few drawings here are just sketches of what expressions I might model.

These are what the blend shapes finally look like.

I then decided to do some colour experiments. I drew these two pictures first as templates and then photocopied them and painted them with watercolours.

For Pocahontas, I just tried different colour variations of hair, clothes and skin. Skin is a very difficult colour to get right, especially when you have only 10 different watercolours. I don't think she looks good with blond or red hair. So brown or black hair. Eyes work if they are black or green. Clothes are best if they are contrasting primaries, so no patterns or mixed tones. She looks nicer pale skinned too. Personally I like the top-left one, which is closest to the colours she is now, but the one right of it, just need to change the eyes to green, and in the bottom left corner of the page could work too.
For Cornelius, I think black eyes work best. The one in the middle row to the left is more of how gorillas are actually coloured, which is a bit boring. Blue I don't think works too well. either. I like my original chocolate colour of course, but I like the strong contrast of the black and light brown of the one next to it. The one in the bottom row to the middle could also work too. For the moment however I think I will keep both models as they are.
Paint-Weighting, I hate Paint-Weighting. It's so fiddly and it relies so much on trial and error which eats up time. Pocahontas was skinned with a Drop-off rate of 0.4 and Cornelius with a Drop-off rate of 0.6. This made Pocahontas' belly deform when she rotated on her hips, moved her arms or her legs, so I tried to go in and reduce the drop-off rate to each joint by paint-weighting. Unfortunately, I think something went wrong, (I probably did it too low) because when she moved, the model just broke up into sharp fragments. I going to have to do it again. I hoped to start animating by next week, so I would like to get this problem done and soon.

No comments: