Two Possible Feature Reaquests for V.4 (Technical Fesiblibly Dendent)

Started by Cyber-Angel, July 23, 2015, 11:09:26 PM

Previous topic - Next topic

Cyber-Angel

I have two requests that I'd like to see implemented in terragen for V.4 or perhaps later development resources allocation,time, application road-map and technical overall feasibility withstanding fall into two camps:

Procedural Variance System for Objects (Artist Interactive) (Vegetation and Trees, mainly):

Description Of Problem:

Populations and the system that creates them are largely the same object repeated (With limited height and rotation variance) that cover the terrain to x coverage area resulting in a pleasing yet samey results (Scenes lack natural variation due to same object beginning repeated to times x).

Proposed System:

A Procedural creation tool (With clearly defined rule set, such as trees needing trunks) that takes any base model (Blueprint, if you will) and with a single click produce say one hundred variations these on a thumbnail type viewer selection of any of which (generated models act as base parent) can produce variations.

This is dependent on how many base models the user has to start with that the system can draw upon (think database of base model parts the system can use) the user after creation can then send any number of these assets to act as members of a population: other population nodes could be created from within the tool whilst the artist continues to create assets!


Feature Request Two:  Implementation Of Johan Gielis Super-Formula for improving Terrain

I have recently come across the work of Johan Gielis the so called Super-Formula as described in the 2003 paper (A generic geometric transformation that unifies a wide range of natural and abstract shapes) if it could be implemented it could dramatically improve the natural appearance of terrains in terragen.

_________________________________________

If the above would be not possible for any reason, thank you in advance for taking the time to look at these ministrations of mine!

Respectfully yours,

Cyber-Angel         



       

Oshyan

Making unique variations of non-procedural models that *look good and fit together* would be *tremendously* difficult unless you had a library of base shapes that were *all explicitly designed to work together in any combination*. Even if you had such a library the requirement for these base shapes to work together in many combinations would place strong limits on flexibility and realism and/or the number of realistic variants. So even in the best case scenario for such an approach, it would take a lot of modeling time, increase the size of the Terragen application by 100s of MBs, and the end result would be mediocre if not downright bad-looking (at least compared to more advanced procedural and hand-modeled plants). A true procedurally-based plant system is by far the best way to achieve these goals. It's not a technology we have in-house at present, but there are many other high quality procedural plant modeling solutions and with the upcoming SDK availability hopefully one of them will choose to make a plugin for Terragen.

I looked at the "superformula" stuff and it really looks more applicable to something like plant seed shapes than terrain. The shapes are almost invariably symmetrical (something rarely seen in real-world terrain), and look more like diatoms than anything. I don't really see the value in using it for terrain, but a node that generated such shapes based on the root formula and variations of it *would* be interesting, if only because it would add another tool to the toolbox. That being said it is, again, probably a better fit for someone to write a plugin to implement it rather than doing so as a core addition. This is the kind of simple, clean, stand-alone generator formula that would be ideal for a plugin author to use as a first project, I think.

- Oshyan

Cyber-Angel

Ah, OK sounds a lot more problematic than I anticipated! The last thing I'd wont is code bloat, with the associated performance issues. What I had in mind with the Johan Gielis Super-Formula was for the kind of terrain found in some parts of Thailand, but anyway I'm sure anther paper would be more applicable somewhere down the road!

Again Thank you for your consideration in thees matters.

Regards to you

Cyber-Angel   

Oshyan

Thanks for the suggestions and ideas. We definitely hope to see some of these more unusual or out-of-the-box ideas being experimented with by SDK developers in the future. No other terrain-focused 3D application that is currently supported even has a proper SDK. There is a lot of potential!

- Oshyan