Test with interior lighting only

[Peter Milner]

I thought I'd compare the 3 engines using only interior lighting (no daylight).

The time is in minutes and seconds.

[Rich Rosemann]

Humm....very nice meeting room model, Peter.

So, engine 1 looks very good, more subdued, strong shadows, nice highlights, looks more complete.

Engine 2 looks flatter, weaker shadows, grainer.

Engine 3 much flatter with almost no shadows.  The leather chairs look very good, though.

 

I'm judging a lot by the shadow depth.  Considering all lighting is located in the ceiling, which, to you, is more accurate?

 

 

[George Ioannidis]

My impression is rather inverse. To me the third image (engine 2) looks more vivid as colours and shadows and materials look better (especially the leather on chairs). First image looks like it has slight grey transparent tint everywhere, mixed in all colours. If I were Peters customer I would choose the third one. It's all my absolutely subjective choice though (may be strengthen on my monitor due to its complete "non calibration" as well :D ). Finally I agree with Rich as regards to the second image, which is too grainy to be comparable with the others.

[Jan Verzelen]

In my opinion, e2 looks way too colorful. Maybe this a monitor thing of course. Engine0 looks like a photograph and e2 looks like a drawing to me. One thing that strikes me is the lack of shininess on the table base in e2. Is the leather supposed to be grey or black?

[Peter Milner]

The leather is black.

[Roy Hirshkowitz]

Great test-- I'll comment a little further when I get a chance.  Just so you know, the overall brightness difference seen in Engine 2 is very likely a direct result of a tone-op change I made (and like.)  I may apply that change to the older engines-- haven't decided yet.

I'm actually working on another engine right now.  The work is going very well.  I may have something as early as next week to show you.  My primary goal for all of this stuff, BTW, has been to eliminate two significant sources of error in the current engines: transparent shadow rays and daylight portals.  The transparent shadow rays is the piece which makes caustics difficult, and the daylight portal piece just introduces a bunch of inaccuracies.  I'm trying to do all of this without increasing calculation time or artifacts-- getting closer.

Not sure if I'm going to retain either of the two current experimental engines-- although e2 is looking the most promising.

[Miguel Tano Adanes]

At first sight, in this case I prefer e0. According to Jan it looks like a photograph, a part the leather of chairs that look more realistic with e2. I suppose that if e1 would have run further, it would have reached excellent results, yes, but in how many hours? e2 still appears too much vivid in my opinion. 

[Marc Chaumier]

I must say, I prefer the e0... The feet of the table, the screen, the spot on the table, the colors not too saturated...

Perhaps the shadows are a little bit too dark...But in artificial lighting, I believe its true. And in the other hand, that give a good contrast. It' true that in the e2, the leather looks very good, due to an higher contrast, perhaps. For me, the ideal would be between the e0 and e2, but closer to the e0.

Good job, Peter.

[George Ioannidis]

For me, the ideal would be between the e0 and e2

Can these images be combined somehow in nXt image editor as we can do with consequent renders? Maybe it could give some interesting result?

[Marc Chaumier]

Peter, is the e0 the packet tracer or the path tracer?

[Peter Milner]

E0 is packet tracer and E1 is Path Tracer.

[Peter Milner]

At the moment I use packet tracer for quick renderings and path tracer for quality renderings.

I like the colour bleed and more subtle shadows of E2, but it does seem to have lost some reflection on the table base.

Also the lighting balance seems to be completely different, particularly with the row of spots at the end (see the door handle shadows).

[Peter Milner]

Here's how I would normally present this image:

Engine 1 - 50 minutes - 420 passes and tweaked in Photoshop.