Getting started

Explains how to install the renderer

Download the distribution

Here we explain how to install and run the lightmetrica renderer using the binary distribution. Another option is to build the framework from the source. In this case, please refer to this section.


In general, rendering is CPU intensive job which can be massively parallelized. In our implementation the most of the renderers utilizes the power of parallelization, so the large number of cores would be recommended. On the other hands, the memory consumption of the renderer depends on the complexity of the scene file. You want to select appropriate hardware configuration suitable for your requirements.

Download and install

Click the Download in the menu bar, and download the distribution for your environment. After download, unzip the file, and you will find the directory containing an executable lightmetrica. The distribution also contains some example scene, and files for the plugin development.

Next open the console (command prompt for Windows) and move to the directory you just extracted. Executing the file lightmetrica would generates the following message. If you can see the message, the setup is complete.

Usage: lightmetrica [subcommand] [options]

Welcome to Lightmetrica!

Lightmetrica: A modern, research-oriented renderer


- lightmetrica help
  Print global help message (this message).

- lightmetrica render
  Render the image.
  `lightmetrica render --help` for more detailed help.

For uninstall lightmetrica, just delete the extracted directory.

Command line options

As the initial help message suggests, executing lightmetrica render --help command would generate the following message:

Usage: lightmetrica render [options]
  --help                        Display help message (this message)
  -s [ --scene ] arg            Scene configuration file
  -o [ --output ] arg (=result) Output image
  -v [ --verbose ]              Adds detailed information on the output

Current options are fairly simple. -s or --scene option specifies the path to the scene configuration file. -o or --output option specifies the name of the output image. Note that we don’t have to specify the extension of the output file (e.g., some_result instead of some_result.hdr) because the output format is selected within the scene configuration file. For enabling debug messages, put -v or --verbose option.

Rendering an example scene

An example scene

An example scene

The binary distribution contains several example scenes in example directory. Let’s render one of the scenes. We assume the current directory is the same directory as the executable lightmetrica is placed. Executing the following command would generate an result.hdr file.

./lightmetrica render -s ./examples/001/scene.yml

As the command implies, we need to specify the scene file with -s option. Also we can change the output file name with -o option. The example scenes generate images with Radiance HDR format. In order to open the file you can use the software like this. If you can successfully open the rendered image, you would be able to see the following image. If something wrong happens, please look at Trouble shooting section.

Reading output messages

We designed the output messages from the renderer as human-readable as possible. Here we describe how to read the output messages including error messages from the renderer. In the first few lines of the log message would be like this:

| INFO  0.001 |
| INFO  0.002 | Lightmetrica
| INFO  0.004 |
| INFO  0.005 | A modern, research-oriented renderer

The output messages are formatted with the same manner. In the first row separated with |, we shows the type of the messages and the time (in seconds) that the message is generated.

If you enabled the -v options, you can see the addtional rows useful for the debugging purpose:

| INFO  0.003 | main.cpp~ | @ 402 | # 0 |
| INFO  0.005 | main.cpp~ | @ 402 | # 0 | Lightmetrica
| INFO  0.008 | main.cpp~ | @ 402 | # 0 |
| INFO  0.011 | main.cpp~ | @ 402 | # 0 | A modern, research-oriented renderer

FILENAME shows the name of source file where the message is generated, LINE is the line of the code where the message is generated in the file specified by FILENAME. TID shows the ID of the threads from which the message is generated. Note that the thread ID changes from run to run, except for the ID of the main thread (0).

There are four types of log messages as follows. The messages are colored according to the severity of the message types.

Type Meaning Color Description
INFO Information notice White Messages for reporting rendering processes. Scrutinizing this type of messages is a good way to find inconsistencies when you have some errors.
WARN Warning message Yellow Messages for something unexpected happens in the renderer, but not affect the execution of the rendering.
ERROR Error message Red Something wrong happens in the renderer. The application should stop immediately.
DEBUG Debug message Gray Messages only for debugging purpose. You can safely ignore the messages.


We deliberately changes one character in examples/001/scene.yml. The sensor_2 are not defined and this file would generates some errors.

  - id: n1
    sensor: sensor_2  # Changed from sensor_1

Executing the command lightmetrica render ./example/001/scene.yml would generate the following messages (partially excerpted):

| INFO  0.152 | ............ Loading asset 'sensor_2'
| ERROR 0.156 | ................ Missing 'sensor_2' node
| ERROR 0.162 | ................ See around line 5 @ scene.yml
| ERROR 0.166 | ................    3 |
| ERROR 0.169 | ................    4 |   assets:
| ERROR 0.174 | ................    5*|     sensor_1:
| ERROR 0.183 | ................    6 |       interface: sensor
| ERROR 0.191 | ................    7 |       type: pinhole
| ERROR 0.197 | ............ Failed to create emitter
| ERROR 0.202 | ............ See around line 84 @ scene.yml
| ERROR 0.206 | ............   82 |     nodes:
| ERROR 0.210 | ............   83 |       - id: n1
| ERROR 0.214 | ............   84*|         sensor: sensor_2
| ERROR 0.218 | ............   85 |         transform:
| ERROR 0.222 | ............   86 |           lookat:

The error messages shows several candidates to check:

  • Missing 'sensor_2' node implies there is no definition for sensor_2 and suggests to see around the definition of assets (line 5).
  • Failed to create emitter implies failure to create the sensor and suggests to check the reference to the sensor_2 (line 84).

In this case, the latter is correct implication to the cause, so we can successfully fix the error by changing sensor_2 to sensor_1.

This example shows the basic process of the debugging the scene configuration file. As we can see, the log message is an powerful tool to find the cause of errors.