I've made a wristwatch model. It consists of many parts, so I've created an instance of finished project, so that I can easily transform (rotate, scale or move as a whole) it for rendering according to my needs.
However, I created the model approximately 4 times larger than a real life wristwatch size. I did so because I was advised earlier when I was a beginner, so that I don't have to deal with some 'issues' while modelling - like zooming on parts, entering very small values in nodes (like bump nodes, noise node). So, it felt more convenient.
But for rendering they advised that it's better to render it at real life scale. That's why I scaled down the instance 4 times smaller.
There's a texture I'm using for the watch's dial (highlighted blue surface): 
Here's an example image texture I'm using:
For testing, I rendered same scene using both approaches:
- Keeping original scale of instance
- Scaling it down 4 times smaller
Note that size of wristwatch was kept almost same in the rendered result by adjusting the camera. So it wasn't like the wrist watch was actually looking 4 times smaller in the render. I hope you'd understand it. Here's a rough illustration to explain it further:
I did not notice significant difference in the quality of that texture, but I suspect it affects the quality of image textures. I'm not sure by my comparison test.
Given all that, is this approach of modelling at large scale and then scaling down the instance significantly, a good or bad practice? Can it lead to quality issues of image textures (or maybe some other issues as well)?
If it creates quality issues, then other than rendering at original scale of instance, are there any ways to optimize texture so that scaling down doesn't affect the render quality noticeably?
