The Corolla is Finished(-ish)
March 12, 2012
Posted by on
Finishing stuff is very important. If you keep having new ideas and starting too much and finishing too little, it can all add up and you can start doubting your own power of execution, even if you’re fully capable to do what you were doing. You just don’t have proof (to yourself) that you can actually finish stuff… which psychologically is pretty powerful. Luckily, the inverse is pretty powerful as well: finish something, then use that momentum to finish something else, and that builds up to a productivity powerhouse.
Looking at the backups of the blend file for the Toyota Corolla I was modeling, I can see I started it on the 22nd of June 2011 (whaaaaat?! How long ago is that?!) and worked on it on and off for a couple of months. Then I stopped, and picked it up for a very brief time in November. Then it started rotting :)
So I picked it up again a couple of weeks ago and promised myself I wouldn’t start any other personal project before I finished this one. Not that easy – every time I go to the grocery store I remind myself to bring the camera next time to take reference pictures of a cool motorbike I want to model, and I already have a set of blueprints of my brother’s Nissan Patrol… heh. But I actually did stick to it. And here is the preliminary result:
Nothing rolls like a Corolla ;) (click for hi-res)
“Preliminary?” Yeah. The car and materials are finished, but as nice as the HDRI and backplates from hdri-locations.com are, that free sample has been tremendously overused… So don’t worry – I have a couple of other renders in mind. This was just to show to the parents who owned the car and are more easily impressed ;) Something else will follow sometime soon.
I posted more intermediate content in the WIP thread at BlenderArtists, so feel free to check that out. Particularly there’s a couple of slightly outdated wire renders there from when I thought I’d finished the model (rendering showed there were actually a bunch of discrepancies :) ) I’ll probably update those on the next post about the car. But here are some lessons learned:
- Finishing! Finishing finishing finishing. It’s all about finishing.
- Get as many reference pictures as you can, as hi-resolution as you can, and as early as you can. It’s funny that I did a lot of work on the car right after I finished the Vehicle Modeling Series (so I had all the know-how), and now that I came back to it after a few months I was looking at parts I’d done back then and thinking “what? That’s not how it looks!” I didn’t get relatively that much more experience meanwhile, so what changed? Well, I had found a new set of pretty hi-res images that let me look at the details and see all the little nuances. I was doing much better modeling work after that. Don’t underestimate the lower-res images though – get everything you can. Every angle is important to figure out proportions and make sure it’s not just that particular car that has a different feature (it happens, if it was modded or restored, or if it has taken a beating after 40 years of use like in my case :P). Oh and find images of individual parts too – Google Images is your friend, single out what you’re looking for.
- Try to control yourself and render as little as possible until you actually get to the rendering stage. Sure, one longer render now and then for motivation, but don’t let it get overboard. I found myself marveling at my own creation a bunch of times, and we’re talking about Luxrender renders… those half-hours could have been spent finishing the thing more quickly. Also, those samples rendered are garbage when they could be useful for a less noisy render later on. Waste of time and CPU cycles! Don’t do it. Don’t be like me :)
- Be organized. Luckily I got into the habit of naming objects (even though the naming structure could make more sense to be honest, but it works), and mid-way through the project I realized how good an idea it is to separate things in layers. I can imagine it would have been chaos otherwise, so I’m keeping that up.
- Non-destructive methods are underrated. Lattices, for example. Before I mainly used them for detailing, but they are a very good way to bend things into shape while keeping the underlying model with sane geometry. You never know when you need to go back and tweak it later (like when you find better images of a part :P). Rigging is a good idea too, and I regret not listening to myself when I thought of using it early on. The windshield wipers are a good example: I could have modeled them straight on the axes and then posed them as I wished. Every time the windshield changed wouldn’t have meant finding pivot points and rotation axes all over again to correct the wipers’ position… I did save a few transforms and vertex groups for this eventually, but that was hardly a nice solution compared to rigging.
- I’m sure there’s more stuff, I’ll let you know if I remember something important.
- This one is more personal: I’m not that fond of texturing… Maybe it’s the fact that this is pretty hi-poly and has some involved parts, but I can’t say I enjoyed unwrapping a few of those things.
I’m planning to do a small video tutorial on how I composited the car onto the backplate. I only found hints towards the method I used, so hopefully it will be of some value. And of course there will be the couple of (original) renders I talked about, so I guess I’ll see you soon :)