Restaurant Viz - Lightning

[tomislavn]

@agentdark45 - Sun in HDR is already aimed through the main doors, just not straight in the middle, it has around 30 degrees offset - you can see it on the soda machine shadow. But I will try that version today :)

@snakebox - I am pretty sure that the plane (portal material) works both ways, no matter of the facing.

@antanas - First of all, thank you very much for you suggestions. Yeah I have a separate portal plane for each window (7 in total on the front doors, but they do not overlap) - it might be wrong :/ so I will try to cover whole doors with one big portal. I am creating portals from the inside maybe a few inches apart from the actual window, so that might be the problem as well. All my glass materials in this scene are already having the "thin" option ticked.

I am open to more suggestions :) it helps a lot! Now back to experimenting!

[romullus]

I do agree about portals, but not necessary so about windows glass. With Corona, i almost never use "thin" glass hack, because i find it unnecessary. Glass with real refraction works very well in most scenarios and look real.
Here, in this image  i used two boxes (4 surfaces total) for window glasses with physical glass material and it doesn't has any negative impact on render speed.

[tomislavn]

I do agree about portals, but not necessary so about windows glass. With Corona, i almost never use "thin" glass hack, because i find it unnecessary. Glass with real refraction works very well in most scenarios and look real.
Here, in this image  i used two boxes (4 surfaces total) for window glasses with physical glass material and it doesn't has any negative impact on render speed.

Yeah I agree on the glass material, I have tried it now without Thin option and it renders the same speed and it kinda looks the same also - but that's because I am using a simple plane for glass not the box like in real-life example I guess. Tnx :)

[antanas]

Sorry to jump on this thread, but quick question to those who know... when you create a plane and add the portal material to it... does it matter which way the plane is facing? like is there an in and out? or does it boost environment to both sides? 

Also seems like intersecting geometry isn't a problem? or do people put them just outside/inside say of windows?

Well from my tests when portals are made outside the window noise goes away faster, about the direction, well here I'm not sure cause every time I did portals I did them with normals facing toward interior cause that just seems logical to an ex v-ray user and more important what would be the point for portals to work for both sides (if I'm wrong, someone please correct me), about portals intersecting geometry - I had read somewhere on the forum what it does not really matter but nevertheless I try to avoid that just in case. Still I'm surely not the most knowledgeable corona user in the world so better to wait till someone preferably Keymaster or Rewalanche decides to elaborate us on that matter further, of course if there is further to it :) 

[antanas]

Well maybe I've done something wrong or that was some specific build bug at the time which was corrected later, but there are the differences which I mentioned :

[antanas]

There is quite considerable difference in sample count + in the one with thin glass I had to lower that same hdri's intensity to match lighting with that one with the solid glass, cause one thing I forgot to mention before is lighting quantity difference coming from windows if those are done this or the other way, but otherwise all is the same - materials, render settings etc. Just don't freak out by those images that interior was heavily wip at that moment (strange I even bothered to save those) so it is quite unfinished/unpolished and no post at all. Nevertheless notice difference in reflections, refractions and the lighting itself and judging from Rewalance's post in this thread https://forum.corona-renderer.com/index.php/topic,135.msg14557.html#msg14557 more specifically this line "If you use regular thick glass with refraction in windows, any ray that passes through it is considered a caustics and your scene will take extremely long time to render..." that behavior is quite explainable though I didn't notice any considerable speed improvements when using thin glass, well maybe some but surely not extreme :) 

[snakebox]

Well maybe I've done something wrong or that was some specific build bug at the time which was corrected later, but there are the differences which I mentioned :

This seems like a pointless comparison, no data on passes, light is clearly different. 

Interesting topic though, personally I have always used the box method until recently after reading around, now trying the single face plane with (Thin) but I don't notice any real difference in rendering, so I guess ill have to setup a side by side test too as it would be really good to know what is happening.

[tomislavn]

A little update.. raw render while working on the materials, kitchen counter side, 88 passes only :) - looking pretty good!

[antanas]

 Hi I think a texture of the wood\wood-looking tile material on the floor is a little bit too small for that purpose and I think tiling could be a little more dense plus it surely needs some randomization, those tiles look like they've been made using floor generator and if so you could easily randomly shift uvs on those using for example this http://www.scriptspot.com/3ds-max/scripts/mass-randomizer nice script. Anything other than that - hmm, its a bit too early stage of the making to advice something :).
 About that which I wrote earlier (reflection dimming with solid geometry glass) - I've done some more tests, more accurate this time and put results in my own thread https://forum.corona-renderer.com/index.php/topic,2361.0.html , well reflection dimming is still there, yet I'm sorry, I was wrong about lighting differences - it really was my mistake.

[tomislavn]

Hi I think a texture of the wood\wood-looking tile material on the floor is a little bit too small for that purpose and I think tiling could be a little more dense plus it surely needs some randomization, those tiles look like they've been made using floor generator and if so you could easily randomly shift uvs on those using for example this http://www.scriptspot.com/3ds-max/scripts/mass-randomizer nice script. Anything other than that - hmm, its a bit too early stage of the making to advice something :).

Hey Antanas, thank you for your comments :) - I am aware of it and as much as I would love to do that, I am running out of time and since this should've been some quick/rough viz of the place with some simple textures, which in the end got switched to a some kind of a semi/photorealistic project - I can't really afford to work more on it. I am rendering the final pictures now and it should be ready for tonight (deadline).

I will definitely try to do some more work on it in my spare time (which I kinda lack though) so I can have this shine in my portfolio though :)

Here is a raw render of the first view - 230 passes, took 2h45min on my i7 4770k machine. Kinda not sure what to do to make this noise go away completely in the ceiling and above the counter (apart from leaving it cook for like 4-5 more hours which I cannot do at this point).

[romullus]

I think, there's something wrong with some materials in your scene. There's too much noise left on walls for 200+ passes.

[tomislavn]

I think, there's something wrong with some materials in your scene. There's too much noise left on walls for 200+ passes.

What could it be? There is some small bump and reflection bitmap (0.04 value though) in the wall material, other then that is pretty much standard corona material with a little bit darker white (grey) color just not to make it clear white. Did I do it wrong?

EDIT: Added 2nd view, notice the noise left on counter glass (thin, simple glass material pure black diffuse, 1.0 reflection white, 1.0 reflection glossiness, 1.0 refraction, 1.0 refraction glossiness) and again the darker wall areas (in shadow mostly)

220 passes, default settings

[antanas]

Oh I know your pain - deadlines+perfectionism+some amount of laziness or distractions suck badly :(
 On the subject - I would suggest this noise is coming from caustics produced by metal surfaces almost directly lit by those lamps check your metal's diffuse colour to be completely black check it's reflection ior to not exeed 20 (or even 10 in some cases) and if this does not help the noise situation then only longer rendertimes will or matid/wirecolor masks plus photoshop plus neatimage or dfine (noise reduction plugins) are your friends to quickly get rid of that noise without rerendering - at least those help me everytime I'm in such situations which are too frequent for my liking but still are almost inevitable :)

[fco3d]

This is looking a lot better, You should try to put some articulation in your stainless Steel counter, and maybe play with some coated metal instead, because now it look very grainy and I don't think more cooking time will make it look better.
Also what value are you using for your black now they look very dead black.

[tomislavn]

This is looking a lot better, You should try to put some articulation in your stainless Steel counter, and maybe play with some coated metal instead, because now it look very grainy and I don't think more cooking time will make it look better.
Also what value are you using for your black now they look very dead black.

Hey fco3d, thank you for your comments. You are right, more cooking didn't help, which I did was I reduced the IOR of the metal surface to 19.50 instead of 25 that it had before, also the reflection glossiness a bit - it kinda helped with the noise. Black is 3,3,3 if I remember correctly but I have used contrast value of 4 in Corona render options so it did make it tad too black.

@antanas - Thank you for the IOR tip :) it helped!

Anyway, this project is finished - I should post new renders in the gallery :)