RTO Test 1

Started by dandelO, December 18, 2009, 06:07:41 PM

Previous topic - Next topic

dandelO

This is really just a test and nothing I'm proud of as a piece, really.

I simply flung a few populations(2 are walli's pines, 2 are different generations of Xfrog fir and one TG internal grass clump pop') with basic constraints onto a procedural terrain and clicked render with full AA slider settings, with customized(non-adaptive sampling).

Now, for a scene with 5 largish populations so closely spaced, in 2.0, I'd still have been waiting this minute for a V'high quality setting on them(I rendered it during last night's sleep).

This was done in under 7 hours with ray traced objects.

'7 HOURS?' I hear you chuckle. Well, let me tell you, my pc is a single cored 1.35ghz processor with only 1gb RAM. 7 hours for this is FANTASTIC, for me at this time, at least. I really would still have been waiting on this rendering in 2.0, I imagine. Possibly it would have finished already but I doubt it.

Thanks to the PS team for this update, I can now feel free to add more object populations to my scenes.
Since I've been using this piece of crap computer I've really had to shy away from anything too render heavy.

Excellent, thank you.

p.s. I also really love the new AA method of rendering - an aliased bucket that's AA'd post render, this is more like it, a lot more sensible. The ONLY drawback to this is that before, if I wanted to do a quick crop to test settings I could safely see the end result without even needing to wait on the entire bucket rendering. This is really a small-cookie, though. I'm very pleased.

Top class, Planetside, you are the best! :)
[attachimg=#]

dandelO

They're not huge populations, by any means, but they are a damn site larger than anything I could have concieved pre-release.

Linda McCarthy

You done good, dandelO!  This trio of pops looks real good, as does the composition.  Linda

dandelO

Thanks, Linda.

Actually, that's a quintet of pop's. There are 5 different objects in the populations here.

Walli's Pine01.tgo and Pine02.tgo.
Xfrog's Grand fir-Adult and Grand fir-Mature.
TG2's grass clump objects make up the fifth.

Anyhow, this, like I said, was just a test and was posted so I could just give my say on how much of an impact these new settings have made for me.

I'll be getting busy soon, I am on holiday for 2 weeks as of Tuesday-next...

Oh, last night I also got to working on populating craters with the shader array node. They are very uniformly spaced because of the shader array's settings but I'm now working on blending the shader array via fractals and it is coming along very nicely. But that's another story, for another day. :)

dandelO

#4
And, damn. I've just realized that I left the object scales at min=1/max=1 for one population that's covering the right hand hill. I usually use min=0.75/max=1.25.

Not to worry.

Henry Blewer

Nice start with this one. I would add a population to the far mountain in the back and some lower clouds.
I have been enjoying the shorter render times, by adding more stuff; to make for long render times.
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

dandelO

Quote from: njeneb on December 18, 2009, 08:30:04 PM
I have been enjoying the shorter render times, by adding more stuff; to make for long render times.

Ha! Yes, that's the ticket.

I won't go back to this, I don't think. Just wanted to see what the result would be.
If I was to go back to it, I'd not add more trees to the far mountains, I'd simply distance shader those mountains out. As soon as I saw it, that was my instant reaction.

Cheers! :)

Matt

#7
Quote from: dandelO on December 18, 2009, 06:07:41 PM
full AA slider settings, with customized(non-adaptive sampling).

This is probably not the fastest way to get that quality. Maybe you could repeat the test with non-customised sampling and see if you notice the difference? It should be a lot faster with adaptive sampling. If you do notice a reduction in quality, you can increase the AA preset, and it will probably still render in less than 7 hours. Adaptive sampling is the usually a good thing.

Matt
Just because milk is white doesn't mean that clouds are made of milk.

dandelO

Thanks, Matt, I might just have a wee fiddle before bed then.

...
...
...

What?

I mean with Terragen, you dirty minded people! :D

dandelO

I actually wasn't using customized sampling. I thought I had.

AA was still 8 but it was default sampling of 64/16. :-[

Anyway, I fiddled around a wee bit, removed the mountains, changed aspect ratio and POV slightly, rescaled some trees, removed the grass clumps and changed the pixel filter from my usual Mitchell Netravali to catmull-rom, and left it to render. 6 hours, for the smaller render height.

I just have to face it, once again, my computer is crap. Still, this is definitely quicker than the old population renderer. I'll leave this test scene now. Onto new things.

[attachimg=#]

Cheers! :)

Henry Blewer

This is a nicer image. It has more 'polish' to it.
I have noticed a new phenomenon, the new AA seems to average the land surfaces in a gourde like shading.  Displacements seem to blend in together more. Sometimes the sharper edges are better to have. Maybe someone can explain how the new AA filters 'filter'? 
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

Henry Blewer

Never mind. I re read the post Matt made about the new render settings. It seems to be more of a phong shading effect. I'll have to make my displacements so that this is taken into account.
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

Matt

njeneb, enabling ray trace objects makes no difference to the terrain.
Just because milk is white doesn't mean that clouds are made of milk.