Hillside villa, clay to final: a full render breakdown with pass by pass notes

This villa took forty hours of modeling and twenty minutes of Photoshop. That ratio is not a joke, it is the whole argument for shooting your passes properly. If your beauty is doing 90 percent of the work, your pass setup is wrong, not your Photoshop skills.

The clay pass is where composition gets locked. If it does not read at clay, no amount of materials will save it. This one was reshot three times. Too much foreground rock in the first take, too little sky in the second, and finally a 21mm lens tilted two degrees, camera placed below the retaining wall so the villa cantilevers over the frame. That final camera decision alone bought a full point of drama that the geometry could never have added on its own.

Beauty pass out of Corona at 5 percent noise. Alongside it I pulled CESSENTIAL_Direct, CESSENTIAL_Indirect, CESSENTIAL_Reflect, CryptoMaterial, and CryptoObject. Z depth as EXR, not JPEG. Eight bit depth is useless for haze work, and I have watched more than one junior artist spend a day fighting banding they could have avoided by ticking one box in the output panel.

In post: cryptomatte to isolate the timber cladding, warm it 8 degrees in hue and boost saturation 12 percent. Same treatment for the pool water but a hue rotation of 6 degrees toward teal, plus a very light Gaussian blur on a separate copy that I overlaid at 25 percent to sell the caustic softness. Z depth into a Camera Raw filter as a mask for atmospheric perspective at 40 percent opacity, weighted more toward the far background so the middle ground stays crisp.

The last five minutes were the ones the client will actually notice. Dodge the top edge of the cantilever, burn the underside of the eaves, and a very small vignette pulled toward warm rather than black. That is how you fake the bounce light Corona already gave you but the tone mapper flattened out. Nothing on this image is invented in post. All of it is amplification of what the render already contained.