Vray 5 - LightMix not working with Irradiance map ?

[nirsul]

Took me days to figure out .
In Beta 1 - It worked but in Beta 2 and the official release - It only works with Brute Force - WHY ?

[JonasNöll]

I think because certain features were not supported or generated unexpecting results when using irradiance maps, so my guess is in order to prevent that it was removed in the final release.
We should be happy that the good old Irradiance map is still included at all, as it seems not very well liked among the developers aparently

[yavor.rubenov]

Hello,

I think because certain features were not supported or generated unexpecting results when using irradiance maps

That's right - LightMix requires Consistent Lighting Render elements and they are not supported with the irradiance map. In Beta 1 we didn't check for that and allowed the LightMix but in most cases it would produce wrong results.

Best regards,
Yavor

[kosso_olli]

We should be happy that the good old Irradiance map is still included at all, as it seems not very well liked among the developers aparently

Because it is a technology that is nearly two decades old and therefore prohibits the implementation of newer features.
It provided decent results back then, but there are better solutions available now.

[JonasNöll]

Because it is a technology that is nearly two decades old and therefore prohibits the implementation of newer features.
It provided decent results back then, but there are better solutions available now.

I understand it from this perspective, but in terms of better solutions available I disagree. If you eliminate IrradianceMap the only available option in V-Ray for primary GI would be Bruteforce which forces me to recalculate my GI during rendertime. There are many situations where it would make perfect sense to pre-calculate your GI and load it from a file and as far as I can tell V-Ray doesn't offer any alternative for that besides IrradianceMap. (I know you can pre-calculate your Lightcache and save that to a file, but good look using that in your primary GI).
But I also understand in this case you could just say "Ok, then just continue using V-Ray 3 and be happy with your IrradianceMap and don't stand in the way of progress" Anyway I'm happy IrradianceMap is still around, and that's my personal view. From what I can tell you mainly seem to work in high-end print and I guess I would have the same opinion in that case. But once you need to render for example 1000 frames 4k camera flythrough with a deadline tomorrow you might be happy for more available alternatives. This one solution fits all problems doesn't really apply and I like V-Ray for its flexibility.

[trinity3d_cg]

I understand it from this perspective, but in terms of better solutions available I disagree. If you eliminate IrradianceMap the only available option in V-Ray for primary GI would be Bruteforce which forces me to recalculate my GI during rendertime. There are many situations where it would make perfect sense to pre-calculate your GI and load it from a file and as far as I can tell V-Ray doesn't offer any alternative for that besides IrradianceMap. (I know you can pre-calculate your Lightcache and save that to a file, but good look using that in your primary GI).
But I also understand in this case you could just say "Ok, then just continue using V-Ray 3 and be happy with your IrradianceMap and don't stand in the way of progress" Anyway I'm happy IrradianceMap is still around, and that's my personal view. From what I can tell you mainly seem to work in high-end print and I guess I would have the same opinion in that case. But once you need to render for example 1000 frames 4k camera flythrough with a deadline tomorrow you might be happy for more available alternatives. This one solution fits all problems doesn't really apply and I like V-Ray for its flexibility.

I agree, it seems with complex interiors (such as our private jet renderings with dozens of light sources and lots of soft reflections) we might still use irradiance maps for rendering, as render times would be 10x longer with brute force. Is it really the intention that we no longer use irradiance maps?

[^Lele^]

We have officially suggested many many times to use the default settings pretty much in their entirety no matter the content of the scene.

We suggest to adjust min AA for thin or fast moving geo, adjust max AA, and adjust noise threshold.
We also, even in the tooltips and by the presets present in v5, suggest to change the LC settings if one is rendering animations rather than stills.

That is the very best way V-Ray can offer you to render *any* scene.
We actively discourage the use of precalculated sequences also for the LC (Vlado, in bold. It's a keeper.), and the use of any of the old camera-based methods, because the speed hit as a consequence of the loss of the adaptive technologies alone would be severe, all the more so with many lights and complex lighting scenarios, over long sequences.
This before issues inherent with the methods creep in and need addressing.

The IRMap had incurable issues hardwired into its model.
The utter inability with fine geometric detail (even at a staggering computational and memory cost, it remains inferior to BF when trying to supersample.), the horrible behavior with moving or deforming geo and changeable lighting, and the endless, endless, endless wrestling of the settings to get it to behave when -really- it was never meant to.
It got old as we added vines, trees, grass, hair and fur, and those 20 million triangles of intricate geometry, a few times per scene.
As the expectations of clients raised the needed quality, the IRMap was left in its own niche: static images which perhaps have more texture detail than geometric, and that perhaps makes very heavy use of emissive materials, and need done very quick.
Very quick, not very good: you only trade noise of a frequency for noise of another: splotches for per-pixel.

I know the IRMap is used to extremely good effect by -at least- one of our highest-profile clients, but it's a small usability niche it has.
It cannot be made to work with a host of present, and future techs, and so is bound to be left ui-less in v5, as a relic for compatibility's sake, and for the few which simply won't want to give it up.
It's by no mistake that no single competitor suggests using the IRMap (for those which have some form of it) over Brute Force.
Many don't even contemplate interpolated primary (or secondary!) methods, due to their proneness to being fiddly at all levels.

The defaults of BF/LC (which now has the handy presets for stills and animations) will handle anything you'll throw at them without issue *the first time*, also enabling the very best of technology V-Ray has to offer, which depends on per-frame LC and/or BF GI to function properly.
If defaults don't work, it's often on us, not the user, to find out why: it's more likely to be a bug than an issue with the settings, and as such it is likely to be handled.
The time gained from troubleshooting and fixing issues, and that gained from enabling the new technologies, will more than make up for the time seemingly lost with the abandoning of the IRMap.

[trinity3d_cg]

We have officially suggested many many times to use the default settings pretty much in their entirety no matter the content of the scene.

Thank you for the information and detailed reply, Lele. I will study it and take this direction for future renders.

best,

Jim

[kosso_olli]

We have officially suggested many many times to use the default settings pretty much in their entirety no matter the content of the scene.


The IRMap had incurable issues hardwired into its model.
It cannot be made to work with a host of present, and future techs, and so is bound to be left ui-less in v5, as a relic for compatibility's sake, and for the few which simply won't want to give it up.