Your Visual Memory Sucks (aka Why Digital Artists Need to Use Reference)
pavla writes:
The Problem with your Visual Memory
When remembering what we saw in the past, we are pulling from our imperfect, limited visual memory. And unless you have spent a focused time studying and analyzing your subject and were able to remember it, it is very likely that you don’t know all the relevant details needed to recreate it. Inevitably, there will be many elements of the full picture that you won’t be able to recall.
This is not an arbitrary imperfection; in fact, visual memory needs to be limited, allowing your brain to be economical with the information it stores by remembering the “gist” of an object while omitting detail . Simply put, remembering everything would be impractical and just clutter up your brain’s RAM.
My visual memory doesn’t suck at all – in fact when i start modelling i stay away from references as long as i can. This way i can extract the basic concept of a thing out of my neurons. I combine this technique with a modelling technique where i use the least possible vertex count and let the subsurf modifier do the smoothing, resulting in very clean design. Most of the time these things start to look as a toy, but that’s fine for me. Realism is only one way to express yourself among many others and thus overrated – if you aren’t a movie film maker.
Agree with most of the statement. I would also recommend to see how capacious memory can be: Kim Jung Gi -> https://www.youtube.com/watch?v=MGbvhyTZXfs
All depends on the experience, the subject etc.
Yes but that’s good for non realism ( and just to a point), you said it yourself, the objects start looking like toys because your visual memory is limited and not just bad.
Even when you want to create something stylized it’s important to have a reference and to actually observe real objects\structures, for example what Kim Jung G gets as a result is something bordering between european comics and manga in terms of realism, much of the details derives from tratteggio or simple shading and good lineart but that works mostly on a 2D plane where a lot of the information is missing and the brain of the viewer fill what’s missing in the picture, in 3D you can screw up much more easily and in real time (or in animations) the object can be observed from every angle at any level of zoom and has to interact with lighting in a believable manner.
In short i agree on the need of a reference, especially in 3D, i challenge any artist to model a gun of any type and to make it realistic and posable (movable trigger, magazine, barrel etc.) in an interactive 3d space without having a reference, which is pretty important given that assets gets reused and placed in different position even in still images.