long render times for noise reduction

[romullus]

I see you have glass panes in windows, can you show material that's assigned to it?

[mediopolis]

here is the glass material.

I am also testing some other scenes from mental ray (our primary renderer)

[Chakib]

Why are you using a Vray HDRI map for the environment?

Shouldn't it be a standard bitmap?

Cheers.

Why he shouldn't use vray hdri map ?

[Coronaut]

Use simple bitmap, adjust white point and black point in HDRI or EXR depends on what you use for map to adjust gamma.
Exposure compensation works same as on DSLR camera, Highlight compression works same as Clamp output(avoiding over burning white points), both should be higher in your case, almost washed out.
Use Lights multiplier samples on default 2.0
Try using 32 PTS with default maximum sample intensity 20.
Absorption on your glass is way too high, try using double sided for refraction mode.
Try using default values first. Avoid using additional lighting in scene, except for Spots, lamps etc.
Try hiding glass on your windows and default values just to see how it goes, troubleshooting scene to see if there is something you should not have inside.

[mediopolis]

Removed all the glass and the bookshelf.. 37min rendertime.

Update... i updated the attachments with a render with the Glass @ 37min

[juang3d]

What's your result with the Corona Benchmark?

Just out of curiosity to have a comparison time, I want to do a similar test and check your speed becasue I think it's a bit slow and I have a different feeling here, I'm not working yet with such complex scenes but my simple interior scenes are fast, so I'm a bit confused with what I see :)

Cheers.

[mediopolis]

a4 scene attached.  2:44 was the time.

I agree. My scene is not that complex, so it's puzzling me also :)

[juang3d]

Your render times in your scene are very weird IMHO, specially after seeing your benchmark results :P

Try to see what happens if you remove all teh textures, leave the materials intact except for the textures, remove them, and remove also the HDR and any light source and start the lighting from scratch.
Also another test could be to put an override material with a simple grey, and try to render it.

About the vRay HDRI, the thing is that that is a Vray map, so maybe it's doing something weird for corona under the hood, so use a standard bitmap (just my opinion, I'm not sure this is interferring at all)
But it's obvious that there is something weird in your scene :P

If you can share your scene (or better, a reduced version of it that suffers the same problem) we can try it, I think you'r scene is rendering too slow for your system.

Cheers,

 

[Juraj]

About the vRay HDRI, the thing is that that is a Vray map, so maybe it's doing something weird for corona under the hood, so use a standard bitmap (just my opinion, I'm not sure this is interferring at all)

This is definitely not it, so far it's been completely sound to use VrayHDRi bitmap loader, which is vastly superior to default MaxBitmap not only through its controls (360degree, various gamma alteration modes,etc..) but also with its re-sampling methods (bilinear,etc.. compared to pyramidal,summed...), although the latter, might not be beneficial in Corona.

In all my works with Corona I used only VrayHDRi. Since I use very hi-res maps (up to 24 000px) I use the resolution multiplier next to light samples even down to 0.1 and lower. I am not sure if this "sIBL" sort of thing provides better noise convergence, I never tested, but I neither experienced any problems at all.

Highlight compression works same as Clamp output(avoiding over burning white points)

It works as tone-mapping, compressing the highlights. Clamp affects dynamic range of the ouput. Two absolutely different things.

Regarding light portals, I also use them very rarely. They help greatly with noise, but only in certain conditions where the interior space has very limited access to environment.
Definitely not a must.

[mediopolis]

Rendering with a standard override mat was fast!  So, I plan to add them one by one, to pinpoint which was causing the issue.

This render was 3:30

[Coronaut]

[/quote]

It works as tone-mapping, compressing the highlights. Clamp affects dynamic range of the ouput. Two absolutely different things.

[/quote]
You are right! Cheers ;)

[juang3d]

This is more in line with what I could expect from your computer :)

Tell us here what material was the problem when you reach to it please.

Cheers.

[mediopolis]

It was the blend materials (wood and wall materials), were the culprits. Just like the performance debugging page said for slow times.  It is pretty amazing how drastic blends slow down corona. Easily by 50% or more when i added blend materials.. Even adding just my bookshelf wood texture added a good bit of time, leaving all other materials with a standard corona material.... lesson learned! ;)  I wish there was a mental ray & maxwell material converter, which is what all our high-end scenes are done in usually.

Can't wait for distributed render so I can see how my render farm of 100 cores would do! 

[racoonart]

I wish there was a mental ray & maxwell material converter, which is what all our high-end scenes are done in usually.

There is one capable of converting Arch&Design mats but not maxwell materials http://forum.corona-renderer.com/index.php/topic,126.0.html

[mediopolis]

most excellent. Didn't realize this same one did mental ray mats also. Thanks Deadclown :)