So impressed with Thea . . .

Bendbox · **Joined:** Mon Feb 14, 2011 1:33 pm **Posts:** 357

notareal wrote:

Bendbox wrote:

It's like, there's no color bleeding with the progressive engine or something.

Color bleeding is connect to the tracing depth and caustics. Set first low and disable later - you will get fairly "flat" result. Naturally there are some differences between engines, but color bleeding should be there, if used sensible settings. Note, if you use shadow catcher feature, it only will receive ambient occlusion type shadows, no bleeding, caustics... But there is other way available soon for better shadow "catching" when using IBL environments viewtopic.php?f=21&t=7427

Hi notareal,

Thank YOU for this tip! I tested this morning and the results are very clear as to what you say. This is a big help and great to learn!

Bendbox · **Joined:** Mon Feb 14, 2011 1:33 pm **Posts:** 357

Solo wrote:

Here are two BSD presets I used.

First is one I use often for studio based renders, it is high supersampling so it took 45 mins:

Attachment:

Preset interior3 Adapt BSD 45m32s.png

The other is one I use interiors which have thick glass or lots of lights, I guess not ideal for this, it took 36 mins:

Attachment:

Preset inside medium 36m07.png

Then I tried TR1, which is may most used render option, I ran it first for 30 mins and then 1 hour, IMO 30 mins was enough.

Attachment:

hacienda TR1 1hr.png

I used a crf for the last render

Hi Solo,

This is very interesting. So, the TR1 engine, in this particular case, took less to time to clear than the BSD versions. I take it however, that you are NOT finding this to be the norm (that the BSD takes longer than the unbiased with Thea)? Why, do you think, that is the case with this image? It's shear simplicity and lack of complex materials maybe?

Bendbox · **Joined:** Mon Feb 14, 2011 1:33 pm **Posts:** 357

sepo wrote:

Hi Bendbox

Would you mind to share Rhino file?
Cheers

Hi Sepo,

Here you go:

http://www.mediafire.com/?rizohaewol7g50h

Bendbox · **Joined:** Mon Feb 14, 2011 1:33 pm **Posts:** 357

19marco23 wrote:

Thanks for the scene. I'm allowed to do a test I also, however, two materials have changed and added some rafters. I fixed the camera view.

Edit TIME WITH ONE EMITTER is 29 minuti no 42. Sorry.

it is possible that two emitter is less than 2 minutes time?

Cool 19marco23, I like what you've done with it!

Solo · **Posted:** Mon May 14, 2012 1:34 pm

Quote:

Hi Solo,

This is very interesting. So, the TR1 engine, in this particular case, took less to time to clear than the BSD versions. I take it however, that you are NOT finding this to be the norm (that the BSD takes longer than the unbiased)? Why, do you think, that is the case with this image? It's shear simplicity and lack of complex materials maybe?

I am assuming that to be the case, the materials are very simple, there is no complicated mats, simple one source lighting, pretty fast for an unbiased engine, equally fast with Thea's default biased engine.

A similar comparison was done at SCF here: http://forums.sketchucation.com/viewtop ... 37#p404837

Bendbox · **Joined:** Mon Feb 14, 2011 1:33 pm **Posts:** 357

Ok, so I messed around with the Adaptive and Progressive BSD engines. I know it's my lack of understanding the Thea BSD engines right now, but I'm really not happy at all with the look of the progressive output in comparison with the Adaptive BSD output. However, in a few tweaks, I was able to get the BSD result below in 4 minutes 6 seconds. There is still a little noise in the shadow areas along the ceiling rafters, but this image compares with the 19 minute TR1 image I got. Very cool!

So, for those of you that DO know the BSD engines well, how do I eliminate the shadow noise without having to up the supersampling for the entire image? This more than doubles the render time and the rest of the image does not need it. I played with the min and max subdivisions under the general settings tab of the mesh light, but this didn't seem to get rid of the noise.

gif hosting

Bendbox · **Joined:** Mon Feb 14, 2011 1:33 pm **Posts:** 357

Solo wrote:

Quote:

Hi Solo,

This is very interesting. So, the TR1 engine, in this particular case, took less to time to clear than the BSD versions. I take it however, that you are NOT finding this to be the norm (that the BSD takes longer than the unbiased)? Why, do you think, that is the case with this image? It's shear simplicity and lack of complex materials maybe?

I am assuming that to be the case, the materials are very simple, there is no complicated mats, simple one source lighting, pretty fast for an unbiased engine, equally fast with Thea's default biased engine.

A similar comparison was done at SCF here: http://forums.sketchucation.com/viewtop ... 37#p404837

Makes sense, I guess one thing to take away from this, is that perhaps it's best to test both BSD and Unbiased modes first if you've got to do a lot of renders of a particular scene. The BSD image I posted above still has noise, so I'd never send this to a client. If I supersample it to get rid of the noise it takes way longer. By the way, the Vray image has some very, very slight noise in that area too.

Rombout · **Posted:** Mon May 14, 2012 3:24 pm

DavePSB wrote:

Rombout wrote:

Putting some extra maps in the meshes.

I'm intrigued but need to ask a stupid question a bit off topic however: Rombout - how are you adding maps. I guess you mean UV maps? What workflow/software do you use to do that? Do you take the mesh out of Thea to another program for mapping? If so, how do you export it from Thea (in what format), and what software would you use?

No i just added maps to the mesh in the other channels, didnt change a thing with the UV mapping. They already have UVs, so i made use of them

sepo · **Posted:** Mon May 14, 2012 6:12 pm

Cheers Bendbox

sepo · **Posted:** Tue May 15, 2012 11:48 am

Bendbox... the textures are missing in your Rhino file....

So impressed with Thea . . .

Who is online