Using CDenoise#
CDenoise is a special command line tool designed to denoise as a batch post-process a sequence of rendered EXR files from the command-line. This tool is very useful if you want to run the denoiser on a subset of machines that are compliant to the hardware requirements of the denoiser.
Note
CDenoise comes with Clarisse installation package and doesn't require any node-locked or floating licenses to run. This may change in the future.
Important
CDenoise is based upon NVIDIA AI Driven denoiser technology and requires NVIDIA GPUs matching the minimum requirements to run properly on the machine.
Hardware Requirements#
CDenoise is based upon NVIDIA OptiX AI-Accelerated Denoiser technology which requires at least a NVIDIA GPU with CUDA compute capability 3.0 Kepler architecture.
Usage#
cdenoise <image> [<start_frame[[:end_frame][%step]]>] [-albedo [<albedo_layer_name>]] [-normal [<normal_layer_name>]] [-strength [denoiser_strength]] [-output <output_file_name>]
Use -help argument to display command line help.
Arguments#
Name | Description |
---|---|
image |
Filename of the EXR image or sequence to denoise. |
If image
is a sequence you then must specify the padding and the frame range such as:
cdenoise my_image_seq.####.exr 1:500
In this example the image sequence is named my_image_seq
with a padding of 4. CDenoise will process the sequence from frame 0001 to 0500
Optional Arguments#
cdnoise
comes with a few optional arguments used to control the application.
albedo#
-albedo [<albedo_layer_name>]
Specify if the CDenoise should use the albedo
AOV to improve the denoiser. By default, if no albedo_layer_name
is specified, CDenoise will look for the AOV named albedo
in the EXR file. In this example, specifying the argument -albedo
tells CDenoise to use the albedo AOV to improve the denoiser output.
cdenoise my_image.exr -albedo
By specifying albedo_layer_name
you can use a custom AOV instead of the default one. In this following example we specified a custom AOV named my_custom_albedo
to be used as albedo AOV for the denoiser.
cdenoise my_image.exr -aldebo my_custom_albedo
frame_range#
start_frame:end_frame[%step]
In the event specified input_file is an image sequence, you must specify the frame range on which you wish to CDenoise to process. Make also sure to specify the padding on the input file such as:
cdenoise my_image_seq.####.exr 1:500
Optionally you can specify the frame step of the frame range such as:
cdenoise my_image_seq.####.exr 1:500%2
In this example CDenoise will process the sequence my_image_seq
from frame my_image_seq.0001.exr
to my_image_seq.0500.exr
skipping each time a single frame (0001, 0003, 0005...)
normal#
-normal [<normal_layer_name>]
Specify if the CDenoise should use the camera normal AOV to improve the denoiser.
Note
Please note that you can't use the camera normal without albedo. You must specify the albedo argument in order to use the camera normal.
By default, if no camera_normal_layer_name is specified, CDenoise will look for the AOV named normal
in the EXR file. In this example, specifying the argument -normal
tells CDenoise to use the albedo AOV to improve the denoiser output.
cdenoise my_image.exr -albedo -normal
By specifying normal_layer_name
you can use a custom AOV instead of the default one. In this following example we specified a custom AOV named my_custom_normal
to be used as camera normal AOV for the denoiser.
cdenoise my_image.exr -aldebo -normal my_custom_normal
output#
-output image_name
Specify the name of the output denoised image. This argument is optional as by default, CDenoise adds the suffix _denoised
to the input image name. For example:
cdenoise my_image.exr -output my_image_final
In this example, CDenoise will output the denoised image as my_image_final.exr
cdenoise my_image_seq.####.exr 1:500 -output my_image_seq_final
In this example, CDenoise will output the denoised sequence as my_image_seq_final_denoised.0001.exr
, my_image_seq_final_denoised.0002.exr
...
strength#
-strength denoiser_strength
Control the strength of the denoiser output. It basically performs an over operation from the denoised output over the input image. When set to 1.0 you get the full output from the denoise image. When set to 0 you get the original input image. In the following example you get 50% of the denoised output blended over the input over.
cdenoise my_image.exr -strength 0.5