Analyze/decoding BSP ?

[Dschaga]

http://jamiecardoso-mentalray.blogsp...oding-bsp.html

Is it possible to have such analyzefunktion in vray too?
I can't help myself, but this false color images looks really more easy to read and understand then an image with the samples as overlay.

(hope it's not a doublepost)

[vlado]

The sample rate render element is quite a different thing, and is not used in the same way. I could get you an image like that, but I myself have not found it particularly useful, because:
a) You have no direct way to influence the results,
b) V-Ray does quite a good job of producing a good tree without much tinkering on part of the user.

Also, in mental ray with BSP2 (which is the default these days), you cannot influence this in any way so these false color images are pretty much useless.

Best regards,
Vlado


Best regards,
Vlado

[Dschaga]

I thought more of false coloring the samples depending on how dense they are .. or what ever we are looking for (BSP, AA/QMC ...)
Vray offers much more settings to tweak the engine then any other renderer, but without good analyze tools it's very hard to understand what's going on.

[vlado]

For the samples, yes - this is already implemented for future builds. My point was that BSP visualization in particular is not very useful.

Best regards,
Vlado

[Dschaga]

Very fine!

btw, how did you render this image?


It's much better readable then the samplecloud and the distance between the samples , but it would still not show the "oversampled" areas, which i always tried to avoid.

[vlado]

I modified the V-Ray code for the "Delone" interpolation a bit. Doesn't the "show samples" option give you more or less the same kind of info regarding the sample density?

Best regards,
Vlado

[Dschaga]

No - it's a complete different "reading experience", because the sampledots are to big and i can't see how many samples are close together/overlaying or how the samples are "flowing" into the corners.

[cecofuli]

In fact, will be useful have the possibility to change the irradiance map dots. Becouse, for example, at 320x240 dots are big in relation to the rendering. At 6000x2000 are very small.
If we can change the dimension, we don't need the "Delone interpolation method"

[vlado]

Well, you can actually turn on a different mode for showing the samples if you type

renderers.current.gi_irradmap_previewMode=1

in the MaxScript listener. It does not work too well with the "multipass" option enabled though.



Best regards,
Vlado

[Dschaga]

Changing the previewmode is kind of cool, but at least this is an example of what this kind of preview is lacking. A beginner can't see that the sampledensity on the ground is too high, but if a subdiveded mesh is representing this plane or colorcoded dots would be orange/green instead of bright green or blue, even a beginner could understand that there is potential to enhance the rendertime. Colorcoded samples are somehow difficult, because vray have to define what color a dot should have and this means that vray have to define by code if the distance between the samples are too high or too low.
A subdivided mesh is a smarter way to visualize if the relation between the modeled mesh and the representing subdivision is logical or to high or to low.

Anyway... It's maybe not a good idea to reduce analyzetools only to GI samples - there is much more possible to controll the speed and quality of the result.
For example, I would like to know where and how much the dark/bright multipliere in color mapping is be used.