Reading 5th TIFF's channel in Python?
Question:
I have some TIFF files that are the output of a Maya renderer, and I need to composite them with some real footage. The TIFF files have 5 channels, rgba + depth channel for compositing. However, all the TIFF loading libraries I have tried seem to discard the 5th layer.
Is there a way to load this with the rest of the image?
Answers:
Use the following approach:
import cv2
image = cv2.imread('yourImage.tiff', cv2.IMREAD_UNCHANGED)
print image.shape
channels = cv2.split(image)
channels[0] # R channel
channels[1] # G channel
channels[2] # B channel
channels[3] # A channel
channels[4] # Z channel
In compositing app like NUKE, you should use a 16-bit or 32-bit OpenEXR file format instead of TIFF. OpenEXR supports up to 1023 render channels what The Foundry NUKE can read in. Read about OpenEXR here.
Z channel (a.k.a. zDepth) is not ideal for compositing as it brings edge artefacts. Use Deep render pass instead (you can store Deep pass in OpenEXR 2.0 and higher). Read about Z pass artefacts here.
In EXR files, you can store a variety of AOVs, such as Deep, Normals, Point Positions, UVs, Ambient Occlusion, Shadows, Disparity, Motion Vectors, etc… Read about Deep compositing here.
I have some TIFF files that are the output of a Maya renderer, and I need to composite them with some real footage. The TIFF files have 5 channels, rgba + depth channel for compositing. However, all the TIFF loading libraries I have tried seem to discard the 5th layer.
Is there a way to load this with the rest of the image?
Use the following approach:
import cv2
image = cv2.imread('yourImage.tiff', cv2.IMREAD_UNCHANGED)
print image.shape
channels = cv2.split(image)
channels[0] # R channel
channels[1] # G channel
channels[2] # B channel
channels[3] # A channel
channels[4] # Z channel
In compositing app like NUKE, you should use a 16-bit or 32-bit OpenEXR file format instead of TIFF. OpenEXR supports up to 1023 render channels what The Foundry NUKE can read in. Read about OpenEXR here.
Z channel (a.k.a. zDepth) is not ideal for compositing as it brings edge artefacts. Use Deep render pass instead (you can store Deep pass in OpenEXR 2.0 and higher). Read about Z pass artefacts here.
In EXR files, you can store a variety of AOVs, such as Deep, Normals, Point Positions, UVs, Ambient Occlusion, Shadows, Disparity, Motion Vectors, etc… Read about Deep compositing here.