-Ok it seems after further testing that tiny gaps in lines only appear in software mode,there is no gaps in Dpix mode.Also I get a more constant edge width in software mode,using Dpix the lines tend to be less of a constant width which somehow gives a more handdrawn feel if this is what user wants.
-Edge mark will disappear if object has a subdivision or multires modifier.
-Using Software unwanted hidden/or backfacing intersections are shown when object is very close from camera.
-Using Dpix is the opposite issue, hidden/or backfacing intersections appears when object is very far from camera.
ok,it seams to get rid of tiny gaps in Software mode I need to calculate feature lines before rendering every single frame.So if you could automate that inside CtrlF12 that would solve that issue.
Oh hello guys! been some time not active on BlenderArtistd but yeah let me tell you what is going on. I’ve been integrating with DarkDefender’s Smooth Contour Modifier, other than this I’ve also implemented some small features such as directional light controlled line thickness and some other small bug fixes.
Ahh haven’t fixed that, sorry…
Oh yeah, I was thinking the same. Currently the still-frame render caches the result for real-time style adjustment, many scenes are so big so it takes very long to recalculate everything, so I leave the user to decide wheter thery should do it. Will make the cache automatically invalid when any scene content is changed then thanks for the report :).
I was busy with school, and If I had time, I will continue further coding. Also: I’ve been planning on some future development, Soon (maybe one or two weeks) I will post the roadmap online, also including plans to integrate it into 2.8’s master. Thanks to all artists and coders who focus on this project and give feedbacks! Stay tuned:)
This together with Grease Pencil should allow Blender to really start hammering away at one of the only areas Lightwave remains strong in (noting how much effort Newtek places in the area of NPR). People were worried about Blender’s NPR abilities dying out, but instead it is now hitting new heights.
Cool, is it based on 1 vector / direct illumination or will it react to multiple lights / GI? Looking forward to what’s in store, after schoolwork is done of course.
I see, thanks for clarifying. For now it looks normals based (no self shadow influence) from the camera gif above. Don’t know if the Shader To RGB concept would work in LANPR, just to put it out there.
What is sure is that is working with EEVEE. And there are pictures in the thread showing that you can composite EEVEE render and LANPR render.
Point that would be interesting to discuss is how easy and efficient will it be to composite these renders.
For freestyle integration, renderer was made first and then, question about how to mix render came after.
With LANPR, we have opportunity to discuss this aspect while renderer is in construction.
I just fixed ortho and many line rendering bugs this morning. LANPR now produces quality lines suitable for blueprints.
This image takes approx. 5 seconds to render on my laptop in software mode.
did you also made the cache automatically invalid when scene content is changed in Software mode?