New Engines

[Roy Hirshkowitz]

Starting a new thread-- the old one is getting a little crowded.

Here's Jan's with e2 at 30 minutes, 300k pixels.  Daylighting only with all settings returned to default.  Obviously glow from the tv is still present.  I fiddled with the floor a little-- very low sharpness levels are undercounting fresnel reflectivity with this engine-- that accounts for the missing reflections in some glossy materials.  So I raised the Sharpness to 0.15 and to compensate lowered the Fresnel to 0.2.  This one still has a bit of mottling on some of the diffuse surfaces, but it's on its way.

Here's a path tracing-- the floor is different and it's only run for 15 minutes so don't try too many direct comparisons.  I haven't a clue why the saturation is so low-- pretty sure this is a bug. The e2 engine certainly has the colors more correct.  The Path Tracer is competitive with the e2 engine in performance for this particular scene-- probably due to the very large window area and many glossy materials.

[Roy Hirshkowitz]

At 1.5 hrs.

[Jan Verzelen]

looks promising. Little artifacts left. I have the impression that something changed on the transparency of the curtains also. What happens to engine 3?

[Roy Hirshkowitz]

I haven't really been impressed with engine 3.  I'm not going to kill it quite yet, but I'm close.

[Jan Verzelen]

I mentioned the lack of saturation before. Turning on the interior lights makes a big difference in this case. If the light of the tv channel is off, why is there still some glow ( decal tagged as light source)

[Roy Hirshkowitz]

The saturation still makes no sense here.  Colors should look much more like the e2 pic.  There may be a bug in the sky saturation code for the PT-- I won't be able to get to this for at least a week.

Re: the tv, I'm guessing there's some material Glow as well-- that has it's own channel.

[Roy Hirshkowitz]

I justed posted build 294 which has the latest e2, the one I've been using for the last few examples.  Test the engine when you can and let me know.  Try to keep the tests simple if possible.  I recommend 300k pixels and reasonable lighting values.

I'll be out of town all of next week but should be able to post here on most days. 

[Jan Verzelen]

not very important for the moment but there's something wrong with region render.

[Jorge Javier Lara Domínguez]

Roy, para que el motor e2 sea eficiente se debe seleccionar los vidrios como delgados y como entrada de luz?

[Roy Hirshkowitz]

I'm not exactly sure what you're asking-- daylight portals are not required or used in the traditional sense (either user-specified or automatically detected)--  however, removing the glass from windows in an interior, or using a material with IOR = 1.0 will likely result in a much slower convergence.

Identifying glass as Thin is not required for efficiency, only accuracy. All light now refracts as it passes through glass.  A thin face with a glass material (IOR > 1.0) that is not tagged as thin will not refract correctly.  For example, the patch of sunlight on the floor will be in the wrong place.

[Jan Verzelen]

A few remarks. This engine becomes a lot faster with the new build.

Decals tagged as light source and using object color don't seem to work properly, first example.

Caustics are much better,  second example.

[kingjin]

All the default Settings, glass and no label thin, E2 rendering engine, more quickly than PT rendering. Make NXT more competitiveness, thank you Roy Hirshkowitz.

[Marc Chaumier]

It seems that, whith the E2, when I put a nXt groundplan tagged alpha channel, the render never start.