For example a client sends you an image of a sofa or a chair you need to model precisely. And it's in perspective. You ask him/her for more detailed images, perhaps even blueprints and he says that it's the only image available. So you have to deal with it.
And when you start (especially if you are a noob like me) you see that's something is off. Even if blockout seemed right further down the line you start noticing that the model is not that precise. And you have to scrap a lot.
So what's your way of working with such images/vague concept art?
Replies
Try Google image search on the picture and see what comes up.
Something look off when you go down the line is usually because you jump right in to hi-poly when your blockout is still too rough
You will very rarely get ortho's of anything. As mentioned above a good blockout is going to be your best approach, nail the basic shapes and move on from there.
I'm not doing furniture for freelance now but I'm getting ready for this. The problem is that I'm already doing basic blockout and still results are subpar for my taste. I think it will come with practice.
What I like to do is figure out the FOV and position of the camera by identifying the right angles in the image. I use a camera in Blender and a background image set to the reference imiage. I usually figure out FOV/cam position with a 3d cube/parallelepipede superimposed over the reference image. Once that's out of the way, figuring out dimensions is just a matter of scaling/positioning your 3d objects to fit right over the reference image while looking through your 3d camera.
I might be making it harder than it needs to be but it works for real photos. I guess another approach might be needed if you're using concept arts.