Interesting settings - Vray (visopts included)

[alpro]

Yes David, I agree with you. I have quite a few interior and exterior scences of this model that I've already rendered with my settings so when I have a chance today or tonight I'm gonna render some with these settings at the same resolution and on the same machine. As far as displacement, I usually don't use it much however wanted to see if I could create some realistic grass for this scene. And as far as patience, it rendered over night while I was sleeping so no big deal, don't think I could of waited that long during the day.




Mike

[EarthMover]

Jerome, I think a lot of people here and also on the asgvis.com forum would be interested in a solidrocks plugin for vray for sketchup. I noticed in your plugin you change the settings not only for different quality selections, but also I noticed that the settings will change based on the size of the output. Can you explain the thinking behind the optimization based on render output size?

Mike, for the blotchiness, based on the output size you rendered at, you can increase the light cache subdivisions to 1800 or so and in the DMC sampler rollout, you can change the subdivision multiplier to something higher. The SR plugin shows a global subdivision multiplier of 6 at that render size. The plugin also changes the min/max for the IM to -4,-2 at that render output size. For grass I usually get good results by painting the grass area with a contrasting color from the rest of the render, then removing it in photoshop using the select by color range option. (ThomThom's blackout script works good as well) CGtextures.com has free high res grass images that you can you can drop in your renders. Then just use a photoshop clone stamp with a really small brush and fake the displacement along the edges of the grass. Just a suggestion....the displacement results you have achieved looked great, but the render time is astronomical.

[subburb]

Jerome, I think a lot of people here and also on the asgvis.com forum would be interested in a solidrocks plugin for vray for sketchup

good news ! i'll be on siggraph on 3-5 august, and try to meet ASGVIS to talk about SR for Sketchup

but also I noticed that the settings will change based on the size of the output. Can you explain the thinking behind the optimization based on render output size?

sure ! i've just followed the Vray rules, and modified them a little bit.

note that all Vray default values are "prepared" for a 640*480 rendersize. So i've considered SR "base" settings for 640*480 render.

But ! There are some values that are tighly tied to the resolution :
Vray manual (http://www.spot3d.com) says :
- When resolution double, LC subdivs needs to be doubled as well to keep the same quality.
- as well, irrmap min and max rate can be lowered by 1 each time resolution double, for example a -4 -1 irrmp at 640*480 is same than -5 -2 at 1280*960, and so on. the detail enhancement radius (in pixels) can be doubled, too, when resolution double.
At least, the AA subdivs can be lowered as well when res double, as each time the resolution double, it's just like an additionnal subdiv.

These resolution dependant settings are a pain to manage, for example for a 1/4 test render and can take huge rendertimes on big resolutions if Vray settings are not modified....

That's Why SR manages all these changing parameters, in a smart way, with an algorithm wich evaluates the number of pixels to render, and adjusting all these parameters for you each time you change the resolution or for fast 1/2 or 1/4 test render.

i've made a LOT of testings while adjusting SR presets and found some interesting settings (not exactly following the manual..)
For example, the SR0.85a default (factory ) presets make these changes, each time the resolution double :
- Lightcache subdivs are multiplied by 1.5
- irrmap min and max rates - 0.6 (to get a smooth slow down)
- detail radius is multiplied by 1.75
- AA subdivs are divided by 1.2 (to be safe)

Of course all these settings are modifiable by Sr users !

Hope it helps

[dalomar]

Subburb...Fernando and I will be at SIGGRAPH, so look forward to meeting you. We'll be at the ChaosGroup booth which is booth num 3119 I believe.

I really like the solid rocks plugin, so it would certainly be a welcome addition to VfSU. Unlike the ChaosGroup, which hasn't really expressed much of an interest in changing their interface, that's probably the biggest request we get. Subsequently the next full version of VfSU/Rhino will have a tiered interface solution with a Basic UI, Intermediate UI, and an Expert UI (Expert being essentially exactly what we have now). The Basic and Intermediate UI will remove a lot of unnecessary parameters and offer some more user friendly ways to adjust settings than having to deal with numbers.

There will be some more resources available in the next full version to do something like Solid Rocks, specifically the ability to add custom UI. I'm not sure how much we'll be offering in terms of scripting support, but hopefully there will be something to offer as we currently don't have anything exposed scripting wise.

In regards to the changing the settings for high res, there isn't the need to double the LC subdivisions. The reason being is that LC is not really tied to resolution in the sense that each sample area will be the same percentage of the final image (assuming screen sample scale) and although that area will cover more pixels, it will not change in regards to the area of the scene that each sample is describing. Therefore, regardless of resolution, there will still be the same number of samples created, and each of those samples will get approximately the same number of subdivisions. When you begin to change the sample size to add more actual samples, then the LC result with the same number of subdivisions will be more noisy and require more subdivisions. I'm actually an advocate of using smaller sample sizes (I typically have mine at about 0.5%) since that allows the LC solution to capture more detail and make better use of adding additional subdivisions.

[subburb]

Hi Damien,
thanks for this LC explanation, i'll make some test to try to reduce even more my lc subdivs when res is raising... it'll help to gain a little bit more time

i'm interested in a SR for Vray for sketchup dev, as it can have an interest for all of us !
i'll start too see how i can do that (perhaps with a little help ?) in begenning of september.

i'll for sure come to botth 3119 to say to you an hello !! (somewhere between 3 and 5 august)

i'll have exactly the same face than my avatar pic and will wear a solidrocks Tshirt .. easy to target

See you on Siggraph !

[EarthMover]

Hi Damien,

i'll for sure come to botth 3119 to say to you an hello !! (somewhere between 3 and 5 august)

i'll have exactly the same face than my avatar pic and will wear a solidrocks Tshirt .. easy to target

See you on Siggraph !

I would hope so Jerome, because if you didn't (have the same face), that would be pretty weird. haha

Exciting news from Damien and a good preview of things to come for V4SU. I think looking at what Jerome has done with SR and bringing him in to advise for the new interface would be a good move for ASGvis. I think he's really created the kind of simplified, yet powerful interface you guys are going for in the new release.

[subburb]

Hi all !

I've found the scrambler to protect code, so now i'm ready to have a try to make a Solidrocks for SU/Vray...

As i'm really newbie on this platform... perhaps someone can give me the first line of code... ?

First thing to do : change via scripting Vray engine values and/or values into the "Vray for sketchup - Render options panel"

For example, in maxscript, when i change via script a vray value, i write something like :
Renderers.current.lightcache_subdivs = 500
This line forces the lightcache subdivs to 500.

Would be very helpful to get this kind of start point

Thanks by advance !

[thomthom]

First thing to do : change via scripting Vray engine values and/or values into the "Vray for sketchup - Render options panel"

It's a no go - at least with the current version. The VfSU data is stored in attributes as binary data in an unknown format. I tried to reverse engineer it - got most of the data for lights. But the main setting data was much harder. And the problem with the VfSU render settings is that it's read only at SU startup, and written upon model save, so even if you managed to decode their binary format, it'd be overwritten when the model is saved.

The only thing which has potential is the lights. As that data is read every time you open the Light Setting window. So you could potentially create a light manager - which is sorely needed.


So that's the bad news.
But a new version is due to be release soon. The core is rewritten and a lot is now in Python. But I've not looked into how easy it is to hack. I have been asking for ways for us scripters to modify VfSU settings - but I'm not sure if this version account for that. This coming release was intended as a pure rewrite of the core to make VfSU compatible with OSX.
But please join me on the ASGVis forum as ask them for more scripting control. It helps my case when more people ask for it, not just me going on and on all the time.