1. For paper you really want to not have thickness (maybe except for a "folded edge") and utilize diffuse transmission, aka translucency. So you get a mix of diffuse and translucency (not more than 100%), certainly not with 100% white (which would be brighter than snow). Learn to use PBR cheat sheets to lookup fairly decent startup values. For the rest of the transmission component, you could mix in some refraction using geometry/incoming as normal (since it's thin geo) and increase the roughness so you get around what the transmission "highlight" is on the real deal. Finally you could mix in glossy using fresnel, but be aware of inverted IORs or just fake the fresnel using "layer weight -> power -> add" to not have to deal with it. 2. Using Nishita sky *ALWAYS* use strength 1.0 as a starting point and adjust exposure instead. This is now the de-facto reference lighting; you scale indoor lighting to match, and you scale HDRs to approximately match (as you don't know their exposure level). I tend to use a clear sky without sun, and my indoor lighting (office lighting) is similar to what outdoor lighting provides fairly close to the windows; outdoor lighting or nighttime with indoor lighting is about the same exposure. Without input from the lighting engineer (luminance maps) all we have are best guesses. Even trying to match lighting engineers fixture picks (and their data sheets) aren't going to produce 100% accurate results anyway. 3. Leave light power at 1W and color at full white, and drive everything using nodes. Those are multiplied, so it's just best using 1 and white unless you have to deal with Eevee - this is a horrible mess in Blender - why? A decent color for indoor would be 2700K-3500K using blackbody color. For strength, who knows - I'm adjusting strength to whatever would be complementary/replacing for skylight without sun with decent sized windows. Indoor lighting is able to compete with daylight, but not the direct sun. 4. Unless you have some high geo fetish, I'd just bake out the indirect light from the lamp and use it as emission geo, and use up/down spots to complement for direct illumination. As well as using a single curve as a base for the whole object and doing the rest in geo-nodes for efficiency. While paper lamps are a rarity for me, I really can't deal with hundreds of high poly lights. Often I'll just fake light fixtures using simple textures or instance lamps in geo nodes according to whatever the ceiling grid/RCP looks like. 5. As for solid walls, I use this only if I *NEED* the walls to be solid. You might in this case. I normally don't, especially for outer walls. To me it's much more convenient to cull backfaces so I can see through the wall without always having to battle the view. You can't do that trick with solid walls. I also have a utility shader thrown into the back of my shader tree that renders all backfaces invisible, which makes it easier to navigate and see through backfaces of walls while in rendered view. 6. That "bevel shadow" is actually kind of important. You'll never have these things going 100% into the floor surface but always leave some kind of shadow gap. In reality it would vary somewhat, and if I use geo nodes to lower ceiling boards and raise floor boards to get that shadow gap effect, note that Blender is behaving weirdly and introduces light leaks in these locations, meaning you'd have to use some camera invisible light blocker objects. Cycles isn't supposed to be subject to light leaks, so I suspect this is a bug. 7. Yes. An increase in sun disk size is always associated with a lowered sun size strength. It's like putting the sun behind a thin sheet of clouds which makes it appear bigger but also dimmer. Don't neglect the other parameters though. 8. And just to be safe, be sure to verify the paper shader actually works wrt light transport and the current configuration. A lot of people will use SSS without realizing there is *NO* light transport, or not including translucency also ensure *NO* light transport without caustics enabled. Don't add sun&sky, and don't open up the top&bottom of the paper lamp until you know the transport is actually working as intended. It's easy to go "this looks cool" when it's not actually working at all. 9. Those brass metal rings aren't connected to anything. Meaning the rest of the lamp is just floating in thin air. Similarly there is no light bulb in there to case a shadow if the light is turned off. Keep in mind this art isn't always just watched by rendering guys, some of us may be involved in actual engineering and spot silly mistakes like this for some angles. You want to create assets that makes sense from any angle and configuration if possible. 10. Bump mapping... The distance reflects the distance in meters from 0 (black) to 1 (white), and musgrave (very dependent on setting) can exceed black-white range by insane amounts. Keep the strength at 1 and adjust the distance only. The best thing is to temporarily use microdisplacement (adaptive subdiv and displacement only), and use whatever you have to use in displacement scale as bump distance, negate bump if scale is negative. Using musgrave as bump, always check what it's limits are and check if they make sense. I have a special normalize visual node group for this, but you can check using >1 and
I really would like to take a trip over to your patrion. Gonna have to watch the vid again to catch your link. I know if I go over to your patrion I'm bound too get 'all your sweet little tips and tricks. As you know, blender is packed with all sorts of ideas and... well a knife can't cut steel.
Hi Derek. Fantastic video as always! I'm not sure about this, but the skin modifier has its own shade smooth option probably because it's the one generating all that geometry. Your mesh is just a bunch of edges, so the mesh's shade smooth is basically just no-op. So it makes sense (to me at least) that there's a separate option. Again, I'm not an expert or a professional, so i may be way off haha.
Love the clear way you explain all of this stuff. Trying to get more fluent in cycles after mostly working in octane and your tutorials help clear a lot up! Question though, at 22:10 is there an advantage to ctrl+b over changing the passepartout opacity to 100?
I am not sure -- functionally I think it's exactly the same, but my guess would be that changing the passepartout to 100 would maybe still be rendering that outer area even if you can't see it.
