How do you define medium poly versus low poly and how does your technique change?

TheDonquixotic

Please forgive my ignorance in this matter. I'm a visual/interaction designer by day who is seriously pursuing VR apps as a side gig and I'm coming into this stuff all very new though willing to learn. 

I took a Udemy course on low poly Blender modeling a learned a lot. 

I wanted to do more organic stuff so a professional high-poly modeler I knew sent me some of his utorials for Zbrush. Zbrush sculpting is cool but hi-poly work that is then baked, and retopoed is a huge technical pipeline overhead and goes beyond my needs for simple VR experiences. 

I look at stuff like 

This: https://sketchfab.com/models/cb8e95b3cbc041cf967360432512a986

Or this:

https://www.behance.net/gallery/16588335/Stone-Golem-Medium-Polycount

Or this:

https://sketchfab.com/models/4b7dd6fb47af462c90d6fcbfbaf65508

Or this:

https://sketchfab.com/models/aad4c32e997f4efc9571294c7e1d0c8c

And I don't see how they're done with the low poly techniques I've learned, nor do they look quite high-poly. 

Is there a specific approach here that I'm missing? A technique, tool, or program that is best suited for this kind of work?

Obviously the most important thing is the person and the work, not the tool or program used, but I feel like I'm probably going about this the wrong way. 

pior

There are countless of different approaches, for about as many different types of results.

The highpoly to lowpoly baking workflow is one of them and has a huge range of applications, but as you noticed already it is quite technically involved and is not necessarily relevant to your needs.

Your best bet for now is to focus on being 100% fluent with core modeling tools (face/edge/vertex editing, extruding faces, merging vertices, and so on), and also, making sure that you have a solid grasp on UV unwrapping. These are the two core skills needed to produce clean and lightweight models similar to the diorama you linked (The Journey). The Totoro scene simply takes it one step further by relying on one level of subdivision being applied to the models to smooth them out.

All that said, if you are using Blender you can very well leverage its sculpting tools in order to intuitively deform low polygon models, with subdivision being applied on top. This is a very smooth and intuitive way to work and might perfectly fit your needs.

You could also decide to work with very strong limitations (like using the absolute least amount of polygons needed to describe a form). This sort of limit is not a technical requirement anymore, but will force you to build very efficient models. Once these skills are acquired you will then be able to transfer them over to more detailed work. In my opinion people who learn that way usually end up producing very clean, efficient and pleasing work.

TheDonquixotic

When using the blender sculpting tools with dynatopo enabled, I found that the meshes got very messy and difficult to unwrap. That's when I reached out for advice and started learned zbrush from my friend. 

Would it be better to turn off dynatopo, and then use a little bit of subdivision smoothing in order to get a more organic look that didn't require too many additional polygons?

pior

Yeah, what I mean by using subdivision was to simply add it as a modifier on top of a regular polygon model, as was done on the Totoro model you pointed out.

My advice would be to not use anything like Dyntopo at the moment, since this will require you to rebuild/resurface the model later, which is what you want to avoid. However it is good for form exploration. But all that said : a solid grasp of core modeling techniques comes first. The sketchfab viewer allows you to see a wireframe view of the models, so one thing you can do to build your skillset is to simply recreate some of the pieces that you would consider appropriate to your needs. The Totoro and Journey scenes are great references, so you will find a lot of your answers right there.

TheDonquixotic

Thanks Pior. Yeah I have been examining their wireframes which is what first sparked my interested. I was impressed that they were able to create such vivid scenes with such limited polygons and stuff. I guess I just need to get back to modeling then!

metajake

@TheDonquixotic Thank you very much for asking the exact same question that I have been searching for.
@pior Thank you very much for your insight on this!