The latest in Eevee/Grease Pencil demos and animations


(Renzatic) #171

I usually find its the other way around. Lights are far brighter in Eevee than they are in Cycles. This is triply true if you’re using volumetrics of any kind.

(Thinking Polygons) #172

This is a tough business, because in reality Cycles is the wrong here. :v:

(Gilles Charbonneau) #173

Yes, on the GT 730, still haven’t got my RMA for my GTX 960 yet, EVGA takes forever to return my new card, cant wait!

Yes, lights are too bright, needs tons of tweaking, takes around 8GB of RAM to render, still, takes only 5 minutes on that 2GB GT 730, which is not too bad IMO!

(English is not my native language) #174

My applause for you, compiling materials and working with probes should be a headache on that card. I hope you have your GTX 960 soon.
I am fighting with specular reflection of some lamps because the shape is very noticeable in glasses/mirrors, mainly called LGT-Fill lamp. For now it is something that I have noticed problematic in Eevee. The problem is even more present when you enable soft shadows.
Also in that scene, the mesh called GEO-Counter.001 that represents the glass of the showcase is in Wire display mode, but it is not shown as in 2.79. This is difficult to distinguish in 2.8.
Another problem is the Displacement node that adds master/Blender 2.8.

But these are minimal problems. Who cares if you can get results like these in just a few seconds (60 seconds to 64 samples on my GTX 960):
(Some minimal color corrections from Darktable to try to match original Cycles image, and I still have corrections to make in some materials)

Thank you very much Blender devs!!!

(Gilles Charbonneau) #175

Made another test, this time I lowered all lamps by half and turned on soft shadows and high bit depth, I turned on contact shadows on all lamps, took 6 minutes 45 seconds on my GT 730.

(dimitribastos) #176

This is already beautiful. Really nice!

(English is not my native language) #177

You look for the Point Lamp that I mentioned in the message above, it is at the end in front of the window. You reduce the Specular value to approximately 0.2 in this lamp. And you move that lamp a little bit up in Z until you do not observe specular reflection in the mirror in front and the left.

(Gilles Charbonneau) #178

I have uploaded my modified scene to my Mediafire account, if anyone cares to have a look!

(Ace Dragon) #179

If Eevee has something similar to the lightpath node, then you can use transparent planes to eliminate the round specular highlight in the window and in the mirror (sort of like a far more sophisticated version of the gobos photographers use).

(English is not my native language) #180

OK thanks. I’m going to research about it.

(Lsscpp) #181

And you can elaborate and prove that, right?

(Thornydre) #182

I think it has been said during the Ask Me Anything with the developpers at the Blender Conference, and apparently they don’t want to change it to keep compatibility with 2.79 files. (or maybe it was in the last Pablo’s stream, don’t remember)

(brecht) #183

The Cycles strength is correct as far as I know, and uses physically based units. We will make Cycles and EEVEE light settings compatible, and this can be done in a backwards compatible way.

(Thinking Polygons) #184

It was said in some of the blender videos recently. I don’t remember which one tho.

(Gilles Charbonneau) #185

An Eevee test with a walking character!

(Hüseyin Demirtaş) #186

can you share project file ?

(Gilles Charbonneau) #187

Its 500 MB plus the textures!

(Hüseyin Demirtaş) #188

oh. ok… :sleepy:

(Zinogg) #189

another one, all rendered in EEVEE.

(CarlG) #190

Wow! Awesome stuff. :+1: