Fix Bands can be used to remove dark or light bands in the luma or chroma channels of video footage. This filter operates in YUV colour space, and can remove bands that are present in one field of one or more frames of the clip.
The screenshots below shows before and after results when using Fix Bands for luma and chroma bands, and illustrates the type of artifact Fix Bands has been designed to detect and fix.
Note that for video footage, these bands typically appear on only one field of each frame. Applying any sort of lossy compression to the original footage is likely to disrupt the scanlines and prevent them from being detected correctly.
When fixing luma bands, Fix Bands can operate with interlaced footage, or when using the Fields, Upper First or Fields, Lower First de-interlacing modes. All other de-interlacing modes should be avoided used because they are likely to corrupt the scanlines and prevent Fix Bands from detecting them correctly. Fixing chroma bands can be done using either interlaced or de-interlaced footage with any de-interlacing mode.
If the bands only appear in occasional frames, Fix Scanline Flicker may also be used to remove them. Additionally, If both dark and light bands appear in alternate fields at the same position in a frame then they may not be fixed correctly. In this case, Fix Scanline Flicker may be employed to correct the artifacts.
The Luma Band Type parameter specifies the type of bands (dark and/or light) that will be fixed in the luma channel. The Length % parameter specifies the minimum number of pixels on a scanline that must pass the detection test for a luma band. Scanlines that pass the minimum length test will be fixed.
For chroma bands, the Chroma Height parameter specifies the maximum height of a chroma band, measured in pixels. The Chroma Length % specifies the minimum number of pixels on a scanline that must pass the detection test for a chroma band, and the Chroma Contrast % parameter specifies the minimum difference in pixel chroma values for detecting a chroma band. Scanlines that pass the minimum length and contrast test will be fixed.
The screenshot below shows before and after results when using Fix Bands to fix Luma bands, and illustrates the type of artifact Fix Bands has been designed to detect and fix. The bright bands visible in one field of each clip have been corrected.
The screenshot below illustrates the type of chroma band that Fix Bands has been designed to fix. In this case, a band is present in one of the chroma channels and appears as a red stripe in the RGB image.
Fix Scanline Flicker can be used to correct small variations in the overall brightness and colour of an entire scanline in video footage. The filter operates in YUV colour space. The screenshot below shows before and after results of correcting scanline flicker, and illustrates the type of flicker that this filter has been designed to fix.
The Channels buttons can be used to specify which of the luma (Y) or chroma (U and V) colour channels contain scanline flicker.
The Range parameter controls the number of frames on either side of the current frame which will be used to identify and correct scanline flickering. Increasing this value may provide a more accurate result but will also increase the time it takes to fix each frame. The default value is 9, meaning the 4 frames before and the 4 frames after the current frame, as well as the current frame will be used to correct flicker.
The Y, U and V Intensity % parameters specify the maximum variation in intensity in each channel that is caused by flicker. Increasing these values will correct larger amounts of flicker. The default value for each channel is 10%. The gang button can be used to control whether the chroma (U and V) intensity values are adjusted together or individually.
The Constant tick-box is used to specify whether the parameters should be held constant over the entire clip or allowed to vary. When parameters vary, their F-Curves are available for editing in the F-Curve editor and Dope Sheet.
The image below shows before and after results when using Fix Scanline Flicker:
Chroma Cleanup can be used to reduce colour fringing in video footage caused by cross-talk between the luma and chroma video signals, and operates in YUV colour space. See below for a before and after screenshot illustrating the type of colour fringing Chroma Cleanup is designed to reduce.
The Luma Threshold % parameter is used to identify bright areas in the image around which colour fringing occurs. Decreasing this value will remove colour fringes from dark and bright areas, and increasing this value will remove colour fringes only from bright areas. The default value is 20%.
The Luma Similarity % parameter affects which nearby pixels are used to calculate the cleaned chroma values. Decreasing this value will mean a wider range of pixels are used to calculate the cleaned chroma value, but at the cost of potentially introducing some colour flattening. The default value is 80%.
The Chroma Contrast % parameter determines which pixels are considered to be colour fringes that require cleaning. Pixels with chroma values above this threshold are assumed to be corrupted when applying the filter. The default value is 40%.
The Filter Width and Filter Height parameters control the size of the area used to calculate the cleaned chroma values. The default value is 4 for each parameter.
The image below shows before and after results when using Chroma Cleanup:
Chroma Re-Sampling can be used to correct artifacts often found in video footage that occur when the chroma information in the image is encoded at a lower resolution than the luma information.
The sampling scheme is specified using a fixed number of luma and chroma samples, such as 4:1:1 which indicates one chroma sample (in two channels) is used for every 4 luma samples.
After specifying the sampling scheme by choosing a preset from the Preset menu, the resolution of the chroma information in the image will be increased automatically, to match the resolution of the luma information (also known as 4:4:4, indicating there are 4 chroma samples in two channels for every 4 luma samples).
Presets for various common video formats are listed in the Preset menu. Presets are available for DV (PAL and NTSC), DVCPRO (PAL), DVCPRO50, DVCPRO HD, Digital Betacam, Betacam SX, MPEG IMX, HDCAM, HDCAM SR and HDV.
An example of chroma re-sampling is shown below, where an image encoded as 4:1:1 (on the right) is re-sampled back to 4:4:4 (on the left). The artifacts caused by the low resolution chroma information are no longer visible in the left-hand image.
Fix Streaks can be used to fix video dropouts that appear as dark or light streaks (sometimes known as "comet tails") in the luma channel. Typically, dropouts can also occur in the chroma channels causing spikes of colour. Fix Streaks works in YUV colour space.
The screenshot below shows before and after results when using Fix Streaks, and illustrates the type of artifact this filter has been designed to detect and fix. In this case, dark streaks in the luminance channel and spikes of colour in the chroma channels.
Note that streaks cauased by signal dropout affects pixels along a single scanline. Applying any sort of lossy compression to the original footage is likely to disrupt these artifacts and prevent them from being detected correctly.
This filter can operate with interlaced footage, or when using the Fields, Upper First or Fields, Lower First de-interlacing modes. All other de-interlacing modes should be avoided used because they are likely to corrupt the dropouts and prevent Fix Streaks from detecting them correctly.
The Luma Streak Type parameter controls whether dark and/or light streaks will be detected.
The Chroma Contrast % parameter specifies the minimum change in the chroma channels that will be detected as a dropout. Decreasing this value will mean more chroma pixels are fixed, but may also increase the number of false positives. The default value is 5%.
The Luma Streak Length parameter specifies the minimum length of dropouts (measured in pixels) in the luma channel that will be fixed. The default value is 10.
The Luma Drop Level % parameter specifies the luminance level that dark streaks must drop below before being detected (or equivalently, the luminance level above which light streaks must rise). Increasing this value will mean more feint streaks are detected, but may also increase the number of false positives. The default value is 30%
The Luma Drop Contrast % parameter specifies the minimum change in luminance that will be fixed for each streak. Decreasing this value will mean more pixels are detected, but might also increase the number of false positives. The default value is 5%.
The image below shows before and after results when using Fix Streaks:
Stabilize is used to remove high frequency fluctuations in frame position throughout the clip.
The Range control specifies how many frames on either side of the current frame will be used to smooth out motion fluctuations. The default value is 11, which corresponds to averaging using 5 preceding frames and 5 following frames.
The Max Offset X and Max Offset Y controls specify the maximum correction (in pixels) in horizontal and vertical direction applied by the filter.
De-Flicker is used to remove high frequency fluctuations in image brightness from the clip. High-frequency flicker varies both temporally (i.e. from frame-to-frame) and spatially (within the image area of a single frame).
The Channels buttons can be used to specify which of the red, green and blue channels should be used in the deflicker calculations.
The Range control specifies how many frames on either side of the current frame will be used to smooth out brightness fluctuations when correcting high-frequency flicker. The default value is 7, which corresponds to using 3 preceding frames and 3 following frames.
The Red, Green and Blue Intensity % parameters specify the maximum amount of intensity change in each channel that should be considered as flicker. The default values are 10%, meaning that any pixel change above 10% should be considered as image motion rather than flicker. If you still see significant flickering in the image then increasing this value may improve the result. If the resulting image contains artefacts that were not present before, the sensitivity value should be reduced.
The Gang RGB button can be used when adjusting the Intensity % values to ensure changes are made to all channels at the same time.
Fix Scratch can be used to fix significant horizontal scratches in a clip that might be persistent from frame to frame. Both, bright and dark scratches can be identified automatically and fixed in any of the red, green and blue channels.
The Detect Channels buttons indicate which of the R/G/B channels will be used to detect scratches. Depending on the amount of noise in the footage, it may be advantageous to detect scratches in specific channels, even though the scratches require fixing in all channels.
The Type parameter controls whether Dark or Light scratches are reomved
The Min Contrast % parameter controls the minimum intensity difference a scratch must have against the background. Decreasing the Min Contrast % parameter will mean that more faint scratches are detected.
The Max Width parameter controls the width of the scratches (the distance between the left and right-hand edge of a scratch). It is important to set this value correctly to ensure that scratches are fixed entirely. If the scratch width is too small, the entire scratch may not be removed from each frame.
Min Height % specifies the minimum height that of a scratch in order to be detected.
The Growth parameter refers to the number of pixels each scratch will be expanded horizontally by before fixing. This can be used to increase the horizontal size of each scratch slightly and improve the quality of the fix.
The Fix Channels buttons indicate which of the R/G/B channels will be used to fix scratches.
Dustbust can be used to quickly remove significant non-persistent bright or dark spots from a clip.
The Type parameter controls whether Dark or Light or Dark and Light spots are removed.
The Channels buttons indicate which of the R/G/B channels will be used to detect spots.
The Range parameter controls how many frames are used to detect spots. For a spot to be detected, it must exist in the current frame and not in the before/after frames.
Dustbust performs motion compensation to be able to detect spots in parts of the image that are undergoing significant motion. This motion compensation can be performed on the full resolution image or on a smaller scale proxy. The Performance parameter controls how the motion compensation is performed. Increasing the accuracy will improve the results of the motion compensation but will also increase the time it takes to detect and fix spots.
The Contrast % parameter controls the minimum intensity difference required to detect a spot. Increasing this value will mean that only very bright (or very dark) spots are detected. Decreasing the threshold too much may result in spots being detected incorrectly.
Dustbust checks each detected area and fix before applying it to the final result. The Reject parameter controls this check. Increasing this parameter will increase the likelihood of a piece of dust to be rejected as a false positive.
The Growth setting can be used to increase the size of the detected area. Please note that increasing the size of a dust area might cause it to be rejected.
De-Noise can be to remove noise from the clip.
Mode controls whether each channel is de-noised separately in each of the red, green and blue channels (RGB), or whether all channels of the image are de-noised together (Grey).
The Gang button applies edits to all R/G/B channel values at the same time when pressed.
Intensity controls how much noise is removed in each of the red, green and blue channels.
Impulse specifies the threshold for detecting impulse noise. Impulse noise differs from normal noise in that it will appear in the clip as occasionalvery bright or very dark pixels of different colours. The Impulse edit boxes indicate the minimum pixel intensity difference between a pixel and its surrounds before the pixel is identified as noisy.
A value of 100% (the default) means that no impulse noise will be removed. Reducing the value to 50% would mean that impulse noise must differ from its surrounds by more than 50% of the maximum image intensity to be removed. Note that as the impulse noise values are reduced towards zero, the time taken to detect and fix impulse noise can increase significantly.
The Noise Level slider can be used to quickly adjust how much noise is removed from the clip. The default value corresponds to removing all noise (as defined by the parameters) from the clip.
The Edge Preservation slider can be used to ensure that significant edges in each image are not adversely affected by noise removal.