Sorry for disclosing the fact that you PMed me about my critics was a bit short to be really helpful. But I agree with you, the few lines I wrote above are not so useful, so here is a longer version.
When I started car modeling seriously, almost 3 years ago, my skills were really low, way behind yours as they're now. I started numerous projects, looks for topics I opened here in the WIP to get a list. And I didn't complete anyone.
My last project was the Ariel Atom II you found here. I did that for a contest, and I found out that I improved very much about patient modeling and paying attention to details. So now I have a different point of view about my modeling process than before. Before is the article I posted more than one year ago on my website :
Pixels or Atoms ?
Now, I work as follow :
- First, I prevent myself from starting a car when just thinking of modeling it. Just because motivation is volatile, so I let myself a couple of weeks to see if I really want to spend some days/weeks/months on that car.
- Second, of course, I search for blueprints and refpics. But now, I focus much more on refpics than on prints, because I realized than prints are useful only at the beginning of the modeling process (that is, the 4th step)
- Three, I force myself to study on photographies the car. Every single detail, every curve, every feature here and there. I've written a check list to make sure I've enough blueprints : it's just a list of parts around the car : headlights, hood, bumper, ... With sufficient level of details to learn the shapes of the cars.
- Four, I draw what some people calls a "spline cage". I draw it without splines, just a mesh with no faces but vertices and edges. I spend a lot of time on it, with a first step to draw the major volumes of the car thanks to prints, than adding every detail I can figure out on refpics.
- Five, I start modeling of the car. But when previously I produced first a simple mesh of the whole car, I now work a part at a time, not moving to another one until I modeled every single detail. Why ? Because with my former method, I always said myself "let this detail apart, you'll check / model it later" and I never did... So, what is worth modeling is worth modeling it accurately.
I think this way of working suits me better. I don't say it's a universal way to achieve nice models, it's just I'm more proud of my new DB9 than I'm of the elder one.
I hope these words can be of use. Apart from them, I've learned some valuable tips :
- Always keep your meshes as simple as possible. Because it will be easier to fix a mistake in a 100 vertices mesh than in a 400 vertices one.
- Keep surface subdivision level as low as possible : usually 2, maximum 3. It will result in faster render times and less memory occupation.
- Render often, a true clay render is really useful. Try to keep a low number of cameras, to make sure you can compare two renders.
- To model hard edges, just add extra vertices row along the edge. It will increase the mesh density when subdivision is on.
- Spend some times turning around your parts in viewport. At least, Blender default lighting is useful to find out bumps. Even if your renders are fast, this move is even faster.
Last word : pay attention to accuracy. It's more simpler to rush on a given detail than to do it accurately. But some here succeeded in teaching me to value accuracy (hey, thanks Arjen !) And it's more enjoyable, I think.
PS : Sorry for typing mistakes and English faults, I don't really have time to re read my post.
And I willing to toss out what I know and learn the right way to model. So maybe i should keep reading car modeling tutorials and practicing and making cleaner smoother surfaces. I must try to not rush to get it done like you said. Thanks again.
Linear Mode
