Render SettingsΒΆ

The render settings include a lot of parameters to globally control your render.


The maximum number of rays to be shot per pixel.


Adaptive Threshold

The noise threshold below which the renderer stops sampling a pixel.

The Adaptive Threshold controls the final pixel noise quantity. Using Adaptive Sampling allows the renderer to shoot more samples where the noise is higher, while shooting less samples where the noise is lower. Compared to non adaptive sampling, this usually results in a uniform noise quality over the image, and much faster renders for a given quality threshold.

The left image uses Samples = 40 and Adaptive Threshold = 0, the right image uses Samples = 1024 and Adaptive Threshold = 0.05, with equivalent render times. You can remark that the non adaptive render has large differences in noise quality, with bokeh being visibly subsampled, a few fireflies and the non blurred part nearly noise free. On the other hand, the adaptive render has a very uniform noise. Achieving this bokeh quality without adaptive sampling would have required 1024 samples per pixels on the whole image, which is a waste of computation for the areas which would not require that amount of samples!

The Confidence AOV shows the noise quantities for both renders (darker areas have less noise, brighter areas have more noise.)

Apart from bokehs, the adaptive error is quite uniform compared to the non adaptive error.

The Heat AOV indicates the number of samples shot per pixel (blue pixels have lower samples, white pixels have the maximum number of samples.)


Adaptive Min Samples

The minimum number of rays to shoot per pixel.

This value controls the minimum number of rays to shot per pixel before the adaptive sampling is evaluated. In some cases, like in very occluded parts of the scene or on caustics, the adaptive sampling can miss some features of the image, and it can be necessary to increase this in order for the sampler to see these features. Rendering in adaptive with buckets and high threshold may exhibit this issue, and you should increase the minimum number of samples.

Increase the minimum samples from 2 to 8 to fix the motion blur broken trail.


Sampling Mode

Switch between progressive and bucketed rendering modes.

  • "Progressive" renders the full image iteratively, enabling a quick preview of the whole image, and progressively refining up to the final quality.
  • "Bucket" renders the image by buckets. Each bucket is sampled up to the final quality until another bucket is rendered.

Note that rendering deep images requires switching to the Bucket mode.



Use Project Settings

Connects the image settings to the project settings, see Preferences > Project Settings.


Width of the image, in pixels.


Height of the image, in pixels.

Aspect Ratio

Aspect ratio of the image.

The aspect ratio set to 1.0, 0.5 and 2.



Write Images

Writes the images of the pass during a local render (CTRL+R or "Local" render). During a batch or a farm render, the images are always written.


File Pattern

Pattern used to create the name of the output file. The pattern can contain $# items which are replaced by the following:

  • $l is the RenderPass path (including prefix and hierarchy)
  • $L is the RenderPass name (excluding prefix and hierarchy)
  • $n is the RenderLayer name
  • $o is the Output name
  • $e is the eye number (1 for the leftmost)
  • $E is the eye name (as 'left' and 'right')
  • $c is the camera name including prefix
  • $C is the camera name excluding prefix
  • $x is the default display extension (exr, tiff, etc.)
  • $f is the frame number ($03f is frame number 0-padded to 3 digits)
File naming


Preview of the final filename in wip mode, for the first output, frame 666 and the eye #1.


Layer Pattern

Pattern used to create the name of the layer inside the file, when supported. When rendering in exr files or supported formats with layering, this value is used to control the name of the produced layers. You can use specifiers like $f, $e, $x, etc, like in $l_$n_$o. The default value is $n_$o. File naming


File Format

The file format of the output image.

  • Tiff : write a tiff file. This format support 8 bits and float pixels. It applies a gzip like lossless compression.
  • Targa : write a Truevision TGA file. This format support the 8 bits pixels. It is not compressed.
  • Png : write a Portable Network Graphcis PNG image. This format support the 8 bits pixels. It applies a gzip like lossless compression.
  • OpenEXR : write an OpenEXR file. This format supports the half float and float pixels, layers and deep images. It includes various compression schemes.
  • OpenEXR/Id : write an OpenEXR/Id file. This format supports the half float pixels. A dedicated reader is needed to read the image in the compositing software. Using this format, one can isolate any object in the image.


Write Deep Image

Write a deep image. A deep image contains multiple pixel colors in depth. Only for the OpenEXR format. Those images can be very large if not carefully compressed. Wikipedia Deep image compositing


Deep Image Compression Ratio

Controls the lossy compression of the deep pixels. 0 means no compression, 1 means maximum compression. In all cases, the result of the flatten image is lossless. Only for the OpenEXR format in Deep Image mode.


Deep Image Merge Flag

Allow to merge the fragments from different objects in the same deep fragment.


Multi Part

If checked, writes the AOVs as OpenEXR 2.0 multi-part files. If not checked, writes the AOVs as channels (compatible with OpenEXR 1.0). Only for the OpenEXR format.


Meta Data

The meta data to insert in the EXR file header. You can add a metadata per line. Here is an exemple of two meta data :


Pixel Format

Pixel format of the image.

  • RGBA : the red, green, blue and alpha channels are written in the file.
  • RGB : the red, green, blue channels are written in the file.
  • A : the alpha channel is written in the file.



Bit depth of the image.

  • 8 bits : the color components are quantized on 256 values. This produces the smallest files, with the smallest color precision. The color values are clamped between [0.0, 1.0]. 8 bits colors should be written using a sRGB gamma.
  • Half float : the components are stored in a half float (16 bits) which is a good choice for the color channels. No clamping. Only for the OpenEXR format. This mode is not precise enough for texture coordinates (st), positions (P) and depth (Z) AOVs, choose Float for those.
  • Float : the components are stored in a float (32 bits) which is the highest quality you can get. No clamping. Only for the OpenEXR and Tiff formats.


Compression method

The OpenEXR compression algorithm.

  • None : no compression.
  • RLE : run length encoding.
  • Zips Lines : zlib compression, one scan line at a time.
  • Zips Blocks : zlib compression, in blocks of 16 scan lines. The default compression method.
  • Piz : piz-based wavelet compression.
  • PXR24 : lossy 24-bit float compression.
  • B44 : lossy 4-by-4 pixel block compression, fixed compression rate.
  • B44A : lossy 4-by-4 pixel block compression, flat fields are compressed more.


File Alpha Mode

Indicate if the colors are written multiplied by the opacity in the file. In all cases, the alpha channel is written with the opacity left untouched.

  • Unpremult/Unassociated Alpha : Write color independently of the opacity.
  • Premult/Associated Alpha : Write color with the opacity multiplied.



Indicate the LUT to apply to the output image. The default value uses the project's gamma. For 8 bits images, you may want to use a sRGB gamma. For a float or a half float image, you may want to save it in the linear color space, i-e a gamma or 1.0.

Backup Project File Name

The name of the backup gproject file to be saved along rendered images. When rendering images in batch or farm, Guerilla can save a file along with each active Layer. This option is useful if you want to keep a track of the setup that is the source of the rendered image. This value accepts $l and $n specifiers. When left empty, no file is saved. File naming


Image Sampling

Time Jitter

If checked, change the jitter pattern at every frame.

Motion Blur

Globally enable or disable the motion blur.

Motion Blur Density

Change the shape of the motion blur.

The motion blur density set to Uniform, Triangle, Leading, Trailing and Gaussian.

Motion Blur Parameter

The parameter of the variable density motion blur.

Depth of Field

Enables compute of the depth of field.

Bokeh Bitmap

Specifies the bokeh luminance bitmap used to control the bokeh distribution.

Pixel Filter


The filter to use to reconstruct the final image.

A test image using the selected filter. Watch the black and white checker for the antialiazing quality, the multi color hairs for the sharpness and the bright sphere for the ringing.

Box Triangle Gaussian Mitchell Custom Mitchell Sinc BSpline Catmull-Rom

Note that rendering deep images requires a FIS filter mode, such as Triangle FIS.

X Width

Width of the filter, in pixels. 1 is the ideal size regardless the filter type. A value larger than 1 will blur the image, a value smaller than 1 will produce a sharper image.

Y Width

Height of the filter, in pixels. 1 is the ideal size regardless the filter type. A value larger than 1 will blur the image, a value smaller than 1 will produce a sharper image.

Clamp Pre Filter

Clamp the sample values. Try [0,10] or [0,5] to remove fireflies. If both values are equal, no clamping is done.

Clamp Post Filter

Clamp the pixel values. If you want to output positions, normals or any other non ranged values, you should reduce the min value to a negative value (like -10000). Set range minimum to 0 to avoid negative values in colors, due to ringing filters, for instance.

Mitchell Parameters

Mitchell filter B and C parameters.


Max Bounces

Global maximum raytracing depth. This value clamps the value contained in raytracing nodes.

Max Bounces set to 0, 1 and 2.


Light Max

Maximum depth for light

This value controls the photon tracing depth globally. Set to 0 to disable bidirectional path tracing. Increase this above 0 to turn on bidirectional path tracing.


Diffuse Bounces

Maximum number of diffuse bounces. This value can be individually overriden in the shader properties.

Diffuse Bounces set to 0, 1 and 2.


Specular Bounces

Maximum number of specular bounces. This value can be individually overriden in the shader properties.

Specular Bounces set to 0, 1 and 2.


Refraction Bounces

Maximum number of refraction bounces. This value can be individually overriden in the shader properties.

Refraction Bounces set to 0, 1 and 2.


Volume Bounces

Maximum number of bounces in volumes. This value can be individually overriden in the shader properties.

Volume Bounces set to 0, 1 and 2.


Shading Blurriness

Control the shading blurriness, in pixel area.

Smaller values mean finer texture lookups but greater memory usage.


Russian Roulette

Control the russian roulette cutoff.

The russian roulette randomly stops the light path propagation, based on the surface reflectivity or transmitivity. This reduces the average light path length, resulting in lower rendering time for the sample amount of samples per pixel, but usually increases the noise.

Left with Russian Roulette set to 0.0 and right set to 1.0 (half rendering time), with fixed 16 samples per pixel and 8 bounces.

While enabling the russian roulette usually requires more samples, this generally proves a good trade off in render time, especially when you increase the bounces limits.

Left with Russian Roulette set to 0.0 and right set to 1.0 (0.6x rendering time), with identical Adaptive Threshold and 8 bounces.

Higher albedos are less affected by the russian roulette, and surfaces with albedo equal or greater than 1 (e.g. pure red or white ...) disable the russian roulette.



Enable spectral rendering

Using spectral rendering allows rendering more accurately certain effects such as dispersive refractions.

Refraction without and with spectral rendering

Note that spectral rendering is not enabled by default on all shaders and most notably, Surface2 > Dispersion must be increased to fully effect the spectral effect in refractions.


Adaptive Luma Threshold

Controls the low luma threshold below which the adaptive stops sampling


Caustic Blur

Controls the caustic paths blurring

Higher caustic blur results in less noisy images, but can alter the shape and intensity of caustics, while lower caustic blur results in more acurrate caustics, at the cost of noise and fireflies.

Caustic blur set to 1.0, 0.2 (1.6x render time) and 0.05 (2.5x render time)


Caustics Type

Controls which light paths are to be considered caustics


Caustic Buffer

Enable the caustic buffer. Rendering caustics using bidirectional path tracing requires caustics generated by light paths to be stored separately in a distinct image buffer. Enable this to benefit from light paths caustics.



Stereo Viewpoints

One Pass Stereo

Enables stereoscopic viewpoints to be rendered in one single pass. If checked, allows to render all viewpoints in one single pass. If unchecked, forces to render all viewpoints separately.

Force Eye

Force the rendering of a single eye.



The denoiser to use. If "None", no denoising. "Default" is an internal experimental denoiser.

Altus Command Line.

The command line used to run the altus denoiser.

Altus Only Files.

Save the file images for Altus but don't run the altus denoising.


Bucket Size X

Width of a rendering bucket, in pixels. Decreasing this value can save some memory during the rendering, but will increase the render time.

Bucket Size Y

Height of a rendering bucket, in pixels. Decreasing this value can save some memory during the rendering, but will increase the render time.

Bucket Order

Changes bucket rendering order.

Adaptive Locality

if 1, a pixel must have as many sample as its neighbor, so no adaptive. 0.5 means at least half of its neighbor, more adaptive, but less continuity in the number of samples. 0 means no continuity at all.

Opacity Threshold

Amount of accumulated opacity to be considered fully opaque. Decreasing this value reduces memory consumption and rendering time for scenes with large number of semi-transparent objects, such as hair or volumes.

Max Opacity Depth

Maximum number of accumulated hits through a ray with opacity. This value controls the maximum number of semi-transparent layers the raytracer will accumulate. Above this count, the ray is considered culled. Increasing this value may resolve early depth culling at the cost of rendering time.

Ray Batch Size

Maximum number of rays to be shaded simultaneously.

Thread Texture Count

Number of texture tiles held by rendering thread. This value controls how many texture tiles are kept in active memory by each thread. Higher values mean faster rendering, but higher memory usage. Note that actual texture memory is not duplicated across threads, but rather forced to stay resident in memory instead of being flushed.


Default Surface Color

Surface color to use when a primitive has no surface shader nor suface color.

Offset Id

Offset the sample id.

Random Strategy

Crappy Random OR nice Sobol


Light Strategy

Strategy for direct lighting Sample the light, the bsdf, both or both with MIS weighting


Use Default Settings

Force recommanded default settings


Max Vertices

Max vertices per path


Max Diffuse Vertices

Max diffuse vertices per camerapath


Max Specular Vertices

Max specular vertices per path


Max Refraction Vertices

Max refraction vertices per path


Max Volume Vertices

Max volume vertices per path


Light Diffuse

Default number of diffuse bounces.


Light Volume

Default number of volume bounces.