It's good to remember that that when Kerkythea xml (KT, TWL) is converted to Thea scene, it is done with some heuristics (as material systems are different and KT does lack many material features available in Thea)... meaning that conversion result might not be optimal and sometimes (specially with more complex materials) you will get better result, if you build the material from the scratch with Thea or at least do some fine tuning.
Yes I agree. This is actually the crux of what I'm trying to achieve. To build an intelligent enough system that it can parse and "create" (read: translate) the mats from tl/kt well enough to produce desirable results. This will enable KT/TL users access to the power of a commercial renderfarm (something they lack now). Given the "ease of use" (although at the expense of more advanced material and other settings unavailable in TL) - there are a large number of frequent TL users. However, TL runs only (Natively within SU) @ 32bit. Thus it is extremely limited in terms of speed. I believe that providing a commercial "farm" solution to TL users would be a hugely popular endeavor. This would ultimately allow someone with very limited hardware resources to construct their models and apply TL settings - then upload that to the "farm". Thus DRAMATICALLY reducing the lead-time, cost and know-how necessary to produce FAST, attractive renders.
Furthermore, as a value-added service I am constructing a "material libarary" that will be housed on the server. If the translation engine has trouble with a particular material, or the end-user is dissatisfied with the results produced, they will be able to select from this material library a suitable material to "replace" the one in their model. The changes would happen during the translation, and the resulting Thea XML file would reflect them. A few of the "value added services" :
1. standardized/optimized Thea Library of Materials (with a simple process to select the materials to update/change in the GUI during upload/job creation)
2. a variety of post-production filters that can be applied in automated fashion to the final result prior to download by the end-user
3. professional (human driven) retouching of images post-production
4. low-res download of "sample" render prior to authorizing job (to ensure final desired results)
5. GUI based region selection and re-render for problem areas (at reduced cost)
6. many more in progress