User:Yiming/Proposal 2018/Week5

GSoC Week5 Status

YimingWu

Things have done

Last week's plan is almost exactly finished, except some are put off, will then implement them when the line style layering system is designed and implemented.

1. Last week's dpix shader has some weird behavior with non-uniform object scale, Fixed now. 2. Multisample display achieved. Seems no need for user selected multisample because all our tests seems to produce fluent framerate at sample==8. will add this feature if there's extensive requirements. 3. Camera distance controlled line style implemented. Light direction contolled one need further design. (e.g. select light object or input a vector or something...) 4. Material seperate lines and edge marks ok. 5. Made render_to_image work, but don't produce result. probably have something to do with multisample.

Screenshots!

Next week

1. I have gathered some suggestions from various places. I will list the major requests from the users, and see what can be implemented first. 2. For coding, Many people have asked for intersection line rendering, and it is also one major feature in LANPR. I think I will then adapt the data structure to achive intersection computation. (and because future works also dependes on it) One thing about accurate intersection calculation is that it should be pre-calculated on cpu side, and then we can send the cached intersection lines to GPU. so when scene have objet/mesh updates then the intersection cache will be re-calculated. This is a time consuming work. But I still remember there's a paper on GPU intersection calculation but can't find it anywhere. If there's anyone happened to be familiar with it please let me know. Thanks 3. Document some code usages. 4. It seems that dot-dash/ wiggle line style are not quite the interest of the user, so I think I will just focus on LineStyle layer deisgn and adapting intersection code next week.

Questions

I have listed questions (and some answers) here in a seperate wiki page. Check here>>>