Models look super flat and thus unrealistic
polycounter lvl 5
I'll post a picture when I get home, but I thought I'd post this and ask for help first. Basically I use reference images, and then start with a 1x1 plane. I then edge extrude and manipulate vertexes to fit into the rough outline for whatever I'm modelling, adding loops along the way. I can do this fine and well, but the problem is I don't know what to do after that; how to transition from the 2d(x,y) of plane modelling to then account for the depth, and curves, etc of 3d(x,y,z). Do you just edge extrude into the Z axis? Whenever I do that it just looks super boxy and unrealistic. Any tips?
Replies
If you had to model a wine glass, not using spline rotations, lofts and similar techniques, using a single picture as a reference, how would you proceed? Is the method you described fast and comfortable? Why? If not, what would be best to do?
I highly suggest you to go ahead and try.
Obviously its not the best as I'm a beginner. I didn't use any spline rotations, lofts, lathes etc. I used a plane, blocked out the outline of the front of the reference glass, duplicated it (so I had 2 planes, one above the other in Z). Scaled the outside vertices towards each other, welded them (so it somewhat resembled a 3d glass, or cup), subdivided and then used soft selection to scale edges, a row at time, outwards for the curve (Looping as I went, and scaling those rows as well). Is there a better way to do this? Anything I can do to improve that particular aspect of it? Looking back the technique I'm using seems clumsy and un-precise, for example the glass isn't even a perfect circle (from the top view). It seems fast, although prone to error (I had to restart twice because of cruddy topology) so it may just be a waste of time. Any alternative modelling techniques I can look into?
Use 2 views, one orthogonal to see if it fits the reference and the other one in perspective, to constantly check how it looks in 3D.
This is called box modeling.
Oh yeah, I know how to do that. I thought you meant try to model the wine glass using the technique I mentioned in the first post; I thought I was constrained to using a plane&edge extrusion modeling for that exercise. My question for box modelling is that because you cannot simply extrude edges (like you can with a plane) how do you add geometry to continue shaping the model? Are you reliant on adding loops and then shaping those? Thanks for all the help btw.
Adding loops is an extremely common occurrence in modeling and the main reason why everybody is concerned with topology and quads.
There are other ways of adding and removing detail on a surface like insets, extrusions, topology redirection.
In short, both techniques can be used in different ways. You should be able to determine which is better for the task at hand (a single model can comprise multiple tasks).
Thanks again.
Always be working AROUND your model, turning it in the perspective viewport to see what shape you're creating. Always have reference to work from. If not of the exact thing you're modeling, of similar things that show you how it might be shaped and assembled. Try modeling a real object you own that you can have in front of you as you work for practice.
Having a full set of blueprints to work off is a huge help when you can manage it, but even then you'll run into issues where things don't line up or make sense between the different views. You have to fall back on other reference images and your own intuition to make sense of the model in those cases. That's why I recommend modeling an object you can have in front of you; so you don't have to interpret 2D reference into 3D in your head and then from your head to a model.