Vrays own walkthrough tutorial is confusing me

[ruebenad]

Hi Vray Helpers

http://www.spot3d.com/vray/help/150S...ials_imap2.htm

I recently followed your tutorial on a walk through and found some aspects of the all be it great tutorial confusing.

I have listed these aspects in the hope you will consider updating the tutorial to help other less cleaver people (such as myself) get in and make animations with your software or maybe tell me where I’m going wrong.

Issue 1) at point 1.4 you instruct the 2nd GI engine to use light cache, then at point 2.1 you instruct setting the primary GI engine to irradiance map, but nothing about the 2nd, this (for me meant) keeping the 2nd set as light cache causing errors. Should it not say that the second GI engine is set to none?

Issue2) 3.1 tick the normals box…where is it? (Using Vray 1.5 R4)

Issue3) When you download the ‘completed scene’ to have a look; the file shows only the primary engine loaded to Irradiance map and the 2nd engine set to none…is this suggesting we don’t load the light cache to the second engine before rendering the final image, if so why did we make a light cache if we don’t use it come render time?

Any input from more able people would be appreciated because, this bit me due to a deadline at work

[DanielBrew]

Hi Rubenad,

I'll have go at explaining the issues you have:

Issue 1: You should keep the secondary GI set to lightcache. Otherwise you would have no secondary illumination in your scene.
Issue 2: I think this means the 'nrm thresh' check box under 'adaptive subdivision image sampler'.
Issue 3: This one is explained in the notes section at the end of the tutuorial. Basically, Vray automatically saves the lighting from the lightcache inside the irradience map so you don't need to load the saved lightcache at rendertime. However if you are using glossy materials and have checked 'use lightcache for glossy rays' you should then set the secondary GI to lightcache with the mode set to 'from file' and use your saved .vrlmap.

Hope that helps,

Dan

[Clifton Santiago]

As Vlado notes in the tutorial, if you calculate the IR together with the LC, the LC is saved in the IR map. This means you can just set primary to IR and put secondary at none. This will look the same as if you loaded the LC map in the secondary slot.

The benefit is your nodes use less memory because they don't have to load the LC separately. LC fly-through maps in world space can get quite large.

The exception to this is if you use LC to accelerate glossies. In that case you must load the LC separately or all your glossy materials will look wrong.

If you're using glossy materials in an animation, LC acceleration is the only way to go. Without acceleration they are too slow, and with Interpolaton acceleration, you will get severe noise flickering from all the glossy materials. This is of course the same reason you pre-calc the IR map. Interpolation is an approximate method of calculation, and will differ from frame to frame on different machines.

[ruebenad]

Hi

Thanks for the very fast response. I appreciated the answer and have a perfect animation as a result.

My file suffered corruption and caused me to look at the tutorial rather than starting again. Now when I follow the tutorial I get the correct results!



All the best

Rueben

[ruebenad]

"Why can't I build a dirigible with my mind" thats a first life question...http://getafirstlife.com/

lol great looking site

Cheers for the anwers all very informative!

R

[Clifton Santiago]

Ha. I was hoping someone might google it.

Yeah, its such a funny site, I had to take a quote from it. I love nerd humor!

[ruebenad]

A new queston

Hi Guys

With this tutorial, in phase 1 we render the light cache every 1nth frame and in phase 2 we render the irradiance map every e.g. 10nth frame.

When writing the Irradiance map does it store the light cache @ every 1nth frame and the irradiance @ every e.g. 10 nth frame?

I am just wondering why we render the light cache at every 1 nth if the irradiance map writes this off because it’s made at every 10nth frame?

Cheers Rueben

[Clifton Santiago]

No, just set it at every 10th frame and render both. LC will automatically do every frame.

LC is stored with IR. If you are not using LC for glossies, you do not need to load it for your render pass. You can put secondary on none.

As Vlado says in the manual:

We are not using glossy reflections in this scene, but if you are using them and not using the Use light cache for glossy rays option at the same time, then you can set the Secondary GI engine to None for the final animation. This is because you won't need the light cache and turning it off will save memory.

[ruebenad]

No, just set it at every 10th frame and render both. LC will automatically do every frame.

LC is stored with IR. If you are not using LC for glossies, you do not need to load it for your render pass. You can put secondary on none.

As Vlado says in the manual:

Hi Clifton

Maybe I did not make myself very clear. I am aware, when not using LC for glossies I set the secondary to none. In this situatation when the second is set to none and the first is set to IR. The IR map is built from a LC of 1nth and a IR of 10nth frames...what happens to the LC's extra frames are they used or lost?

Thanks Rueben

[Clifton Santiago]

It doesn't matter. LC is calculated (I believe) by evenly distributed sampling across the specified frame range along a given camera path. Like IR, I don't think there is a "per frame" specifically. With IR for example, the results from any given frame are added to those from all previous frames, so the IR map just gets more complete.

[joconnell]

Heya Rueben,

On of the main differences between the primary ir bounce and the secondary LC is the amount of times the light bounce for each. In terms of lighting in vray, you have direct lighting, the primary bounce of GI and then the secondary bounces. The direct light is the light coming directly from your light sources and hitting objects in your scene, much like standard lights in 3d. The primary bounce has a lot of sharp detail in it so it uses very high quality methods like irradiance mapping and brute force whereas the secondary bounces only need to be ambient and smooth.

In terms of your question the big difference in the irmap nth frame and lc every frame is the amount of bounces. If you have a long corridor with direct light hitting the floor and render your primary bounce, the light is only bouncing once so the light won't reach to the end of the corridor. The light cache on the other hand has infinite bounces so the light will keep going until it fills the entire scene. Since the irmap is calculated from the cameras view, if you had one frame in a hall and then turned around the corner, the irmaps single bounce mightn't have spilled around the corner to light it up yet whereas the infinite bounces of the lc map will make sure that the light will fill an entire interior - regardless of which frame you render at. It's all to do with how each of the maps are calculated.

Also of note is that the every 10th frame is an arbitrary number - that's really related to how quick your camera moves. If you've got a camera doing a slow pan over 90 degrees where the end point isn't that different from the start point, you probably won't have to calculate every 10th frame, you just have to make sure that the irmap is sampling your entire scene and not missing anything. have a look at this second tutorial for a better idea what I mean - http://www.spot3d.com/vray/help/150S...ials_imap1.htm

[victor.nsy]

Maybe another way to look at it ... samples information collect has nothing
to do with time. The reason why we have to do every 10th frame for irrad
is due to the fact that your camera or object moved in the animation. If your
camera moves really slowly you can render every 30 frame as it cover lesser area
in predetermined time frame. We need to collect all the samples which will be visible
in our viewport as our camera motion takes place.

The resulting calculation will be used to inform vray at rendering stage how
much of the light is reflected off the viewport visible surfaces ... nothing to
do with time.