Lumini...

[vneves]

Hi, it´s my second render in Corona...

10min.... 200MSI..

In this study I needed  a (exclude object) in the lights. In Vray you can select objects that are not influenced by light. would be a good implementation.

[Ludvik Koutny]

It looks adorable. But i do not understand the hype about raising MSI lately. It will give you little bit more accuracy at the expense of significant increase of rendertime. As long as you use corona in  production environment, you will most often want best speed/quality performance ratio, not best qualit only. This is prime example of a scene where raising MSI should make only very little difference.

[Juraj]

It looks adorable. But i do not understand the hype about raising MSI lately. It will give you little bit more accuracy at the expense of significant increase of rendertime. As long as you use corona in  production environment, you will most often want best speed/quality performance ratio, not best qualit only. This is prime example of a scene where raising MSI should make only very little difference.

I didn't inted to start any "hype" and just waited until someone called it out when it gets misinterpretted.. well it did fast :- ). In my scenes, the difference in GI and highlights is quite big, and render times actually aren't affected much at all. I too agree it won't make sense to just keep such high value default (even though this is how it's set in every unbiased renderer and it's hardly worthless). The default 20 is just too small for most scenes imho and forces users to go for unrealistically high reflective values in exchange, which don't converge to same result (it will still not have nice highlights). Isn't this marketed as photorealistic renderer foremost ? So why the discouraging for going after quality ?

With my tests, the result is exact opposite, the speed difference isn't much worse, and accuracy is quite a different. The problem is a point where fireflies appear, the sampling isn't so worse. Vlado said the same about GI caustics (which are unrenderable in Vray), but just look at recent iRay renders (early versions had cutoff by default and they reintroduced it as "feature"...), where it is the "hype" contributing to absolutely stunning realism (example in Delta Tracing works). Any interior rendered below 200 without serious post-production just doesn't look right to me.

"Production environment" (such a meaningless buzz-word in CGI world, thrown left and right..) doesn't mean effectivness for all, for me it is purely technical quality at any sort of cost (talking only about commercial projects). Lot of studios value quality above all, as do clients.


BTW, sorry, it's very cool image :- ) !

[kumodot]

I am a witness Juraj was the Firestarter. ksrksrsrr   lol
    Just kiding. :)

    Vey nice image  !

     Suggestion to Ondra. A way to show our nacionality near the member name... :)Vneves is a Brazillian too. :)

[Juraj]

     Suggestion to Ondra. A way to show our nacionality near the member name... :)Vneves is a Brazillian too. :)

I want to be from the Republic of Internet :- (

[vkiuru]

Hi vneves, that looks good :) Maybe you could sharpen the bump map of the red metal up a bit? maybe try a larger resolution bitmap since it looks a bit pixelated. Did you model that yourself BTW?

It looks adorable. But i do not understand the hype about raising MSI lately. It will give you little bit more accuracy at the expense of significant increase of rendertime. As long as you use corona in  production environment, you will most often want best speed/quality performance ratio, not best qualit only. This is prime example of a scene where raising MSI should make only very little difference.

The default 20 is just too small for most scenes imho and forces users to go for unrealistically high reflective values in exchange, which don't converge to same result (it will still not have nice highlights). Isn't this marketed as photorealistic renderer foremost ? So why the discouraging for going after quality ?

I thought Juraj made a good point with his MSI tests. I for one found it interesting and fail to see how reaching for higher quality should not be a point of discussion if one has the time for it. That said, yeah maybe this scene doesn't need that kind of a setup but why not let everyone experiment and decide for themselves?

[Ludvik Koutny]

Hi vneves, that looks good :) Maybe you could sharpen the bump map of the red metal up a bit? maybe try a larger resolution bitmap since it looks a bit pixelated. Did you model that yourself BTW?

It looks adorable. But i do not understand the hype about raising MSI lately. It will give you little bit more accuracy at the expense of significant increase of rendertime. As long as you use corona in  production environment, you will most often want best speed/quality performance ratio, not best qualit only. This is prime example of a scene where raising MSI should make only very little difference.

The default 20 is just too small for most scenes imho and forces users to go for unrealistically high reflective values in exchange, which don't converge to same result (it will still not have nice highlights). Isn't this marketed as photorealistic renderer foremost ? So why the discouraging for going after quality ?

I thought Juraj made a good point with his MSI tests. I for one found it interesting and fail to see how reaching for higher quality should not be a point of discussion if one has the time for it. That said, yeah maybe this scene doesn't need that kind of a setup but why not let everyone experiment and decide for themselves?

I am not saying you should not experiment. Where/when did i say that? I am just saying that in majority of cases, it will slow down things a bit.  There may be also minority of scenes, where it will not make much of a difference in rendertime, but that makes sense, if you tend to render one frame for over 10 hours. Usually, you do not have such a luxury.

If you do such extremely radical fakes, like excluding certain object from certain lights, then i absolutely do not understand why would you want to chase for physical accuracy at all by raising MSI.

[vkiuru]

Hi vneves, that looks good :) Maybe you could sharpen the bump map of the red metal up a bit? maybe try a larger resolution bitmap since it looks a bit pixelated. Did you model that yourself BTW?

It looks adorable. But i do not understand the hype about raising MSI lately. It will give you little bit more accuracy at the expense of significant increase of rendertime. As long as you use corona in  production environment, you will most often want best speed/quality performance ratio, not best qualit only. This is prime example of a scene where raising MSI should make only very little difference.

The default 20 is just too small for most scenes imho and forces users to go for unrealistically high reflective values in exchange, which don't converge to same result (it will still not have nice highlights). Isn't this marketed as photorealistic renderer foremost ? So why the discouraging for going after quality ?

I thought Juraj made a good point with his MSI tests. I for one found it interesting and fail to see how reaching for higher quality should not be a point of discussion if one has the time for it. That said, yeah maybe this scene doesn't need that kind of a setup but why not let everyone experiment and decide for themselves?

I am not saying you should not experiment. Where/when did i say that? I am just saying that in majority of cases, it will slow down things a bit.  There may be also minority of scenes, where it will not make much of a difference in rendertime, but that makes sense, if you tend to render one frame for over 10 hours. Usually, you do not have such a luxury.
akes, like excluding certain object from certain lights, then i absolutely do not understand why would you want to chase for physical accuracy at all by raising MSI.

Right, you didn't specifically say we shouldn't experiment, sorry. MSI and its effect on rendertime/quality is still an interesting topic though.

If you do such extremely radical fakes, like excluding certain object from certain lights, then i absolutely do not understand why would you want to chase for physical accuracy at all by raising MSI.

I'm not sure what you mean by "radical". It's pretty common to exclude lights from casting light onto specific objects so they don't create burnt out areas such as, for example, pendant lights might create on a ceiling. Why should it mean you can't chase an accurate result overall?

[Ludvik Koutny]

I'm not sure what you mean by "radical". It's pretty common to exclude lights from casting light onto specific objects so they don't create burnt out areas such as, for example, pendant lights might create on a ceiling. Why should it mean you can't chase an accurate result overall?

Because that itself is such a radical drop in accuracy, that some MSI clamping impact is tiny in comparison. If this is your scene, then try 20 vs 200 MSI, i am really interested in results. It made such a big difference in Juraj's scene probably because some secondary specular interaction (reflected refraction of HDRI or something) which is rather specific case...

[vneves]

the important thing is getting good .. and no matter whether it is correct or not physically delete lights ... is almost the same as in photos using photoshop, because it will use? will be equal fake.