Render passes

When a 3d programme such as Maya or 3D Studio Max renders an image, sometimes the results are not quite perfect. The lighting may be off, the colour may be overblown or something may need changing that would require the entire scene to be re-rendered.

There is a solution however in render passes. Essentially a render pass is used to separate the colourmap/shadows/light/specular/diffuse and other effects onto their own separate layer. Rather than the 3D programme combining these effects, tehy can be layered on top of each other in a compositing programme such as Nuke/After Effects/Premier/ and Sony Vegas. This allows for instance, control over the darkness or hue of shadows on their own layer. From here the shadow could be blurred, sharpened or have other effects that cannot be simulated in the 3D programme it was rendered in.

Hypothetically, a scene that has a bright red light in it was passed to the director and he did not like it. Rather than re-render the entire scene the layer that the light was applied to could be tweaked to a more suitable colour. The compositor already has the information and render pass applied in a sequence, so they can adjust it in real-time rather than waiting for it to be processed all over again.

Render passes always come down to control over several elements on layers, rather than one whole element.

An example of separated render passes, each with their own layer then processed and edited on a single image.

Putting 3D into a 2D photo

When applying a 3D model into a 2D photo, there are a few things to consider in order to achieve a look that ensures the object appears to belong to the photograph:

For starters, knowing the height of objects in the world. By measuring things close to where the 3D model will stand, you can get an understanding of the size of your model and how it should be scaled in the world.

Taking reference photos of objects around where the model will stand (for instance a light post or a building) will give you an idea of where a fake object can be placed to mimic things like shadows.

Knowing where the lights are in the scene and which direction they come from is also helpful, as extra light may be needed in the scene to highlight the character/object more.

Environment ball/reflection capture is an accurate way of simulating the light/shadows/colour an environment and the surroundings reflect onto your model.

An example of an environment map.

Adding fake props/ground into the 3d world to use as dummies to capture light/shadow are also very useful. The object/character may be casting a shadow or refracting light onto the scene, but the photo does not understand that the shadow and light should be cast. So the shadow is baked into models and faked

 by turning the shadow into a 2D image that can be applied over the image.

Some of these effects can be faked by just creating the shadows/reflections and light in programmes like Photoshop and adjusted on an alpha channel. If the programme cannot do a particular task, creating your own effects on new layers may be necessary.

HRD Imaging

HDR Imaging,  (High Dynamic Range Imaging) is used to allow multiple light levels in a single image composed of 3 or more images.

Essentially with a 3 photo shoot, 1 photo would be taken at a darker level and shadow detail, the next at a regular level to capture the detail and the final at a higher exposure to capture the light detail. Capturing higher light level and darker images removes some of the detail if captured, but when applied in separate layers to a regular photo with all detail this allows the image to retain the brightness and dark levels while still keeping the original pictures detail.

Here is an example of a typical 3 level HDR image. The top left picture being the high light image, the center the regular level and the dark to the right. The lower left is a combination of the 3 top images sampling all 3 to crate one detailed image with the high and low level captures blended in.