Daily Builds 1.7

[PROH]

....hmm... not happy with this this move too :(

[Ludvik Koutny]

EDIT:

Yep, sorry to see it go. It was useful in many cases.

[alexyork]

Ruh roh... it's #roundedcornergate again....

But seriously though, if we can once again have the option, that's clearly the best situation for users.

[pokoy]

As far as I'm concerned, I'll not be able to render a specific project and will need to switch to 1.6 whenever I need to render certain assets. Needless to say that's a major pain in the pants.
Please ask us if you plan to remove features users rely on, we've been through this a few times now and it's not a good way to handle changes.

[maru]

Sorry for that move, but the issue is just more complex than it seems. We just had to choose the lesser evil. We are also looking into bringing rounded corners for multiple objects back, so stay tuned! Our developers are known to accomplish impossible tasks. ;)

[Ondra]

I agree that the situation is a bit messy, but if everything goes well, we will have new build with different objects rounded edges working in about 14 days...

[shadowman]

hope to see some news about htis old problem.. 😅

cleaning this as I am moving all similar reports to our internal tracking system. We will give it a shot for 1.7

it is because the reflections becomes partially classified as diffuse when lowering glossiness. Currently there is no easy fix for this

[Ondra]

sorry, we won't do that in 1.7, it will remain one of top priority issues for later releases

[thilima]

I was comparing the two latest builds - 10/07  x  24/07

The 24/07 is cleaning the noise faster but this new improved random sampler doesn´t seems to be recognized diferently by the "noise level limit"


If I render some scene by Time limit eg.10mins.  The 24/07 build will have less noise in the same amount of time.

BUT

If I render the scene using "noise level limit" (5% for example). The 24/07 build will be slower (like 20% slower) than 10/07  to reach the same noise level. But the 5% noiselevel on 24/07 will be cleaner than 5% noiselevel on 10/07.

Don´t know why :/

[Ondra]

The 24/07 is cleaning the noise faster but this new improved random sampler doesn´t seems to be recognized diferently by the "noise level limit"

Yes, this is known limitation. The noise estimation has some limitations, such as not picking this visually recognizible difference. The solution is to start using higher noise limits than before ;)

[romullus]

Standards of noise have been increased recently :]

[Monkeybrother]

I'm very happy with the way Corona is heading, but changing functions that people rely on several times creates a bit of uncertainty. This is a sincere question, not trying to troll you: why release a hotfix for 1.6 with the rounded edges changes (that for many users is a step backwards) if you're planning on changing them again in two weeks?

[pokoy]

I agree that the situation is a bit messy, but if everything goes well, we will have new build with different objects rounded edges working in about 14 days...

Ok, thanks for considering this.
Maybe now is the best time to request a useful addition to the way it behaves: could you add optional direction for foreign/self object? Right now, self object produces negative direction while foreign object produces positive - could we have a dropdown for each so all output combinations are possible? That's the only thing I've been missing.

[ihabkal]

Hmm I observed ram usage increase from 28-30gigs of ram up to 42gigs on that scene i'm working on - not a big deal but seems to be on 15%++ side of things yet speed increase is surely worth it )) Anyway that scene of mine uses some proxies, some forestpack objects and some displacement - well the usual wip mess of a scene so it's hard to tell what can be chewing that additional ram + one strange thing I observed while doing some 4k preview rendering of it - scene started to render using some amount of ram then it did displacement + secondary gi and went up to 42gigs (well more or less the usual thing) then it rendered to around 4 passes (around 10 min or something) using those 42gig and then, suddenly, while rendering 5th pass it jumped down to 36gigs - never seen Corona behave like this - maybe it's something wrong on my end and that initial bigger than +15% ram consumption increase difference I'm observing is not normal after all.

Corona has always had a higher usage of ram at the beginning then it tapers down when it starts making passes.

[steyin]

That speed increase for 5-15% more RAM tradeoff might be very troublesome, especially in an office environment where IT/ownership won't spend the $ for more RAM for several computers.