Frequently Asked Questions

Here are a few topics that may address you questions or problems:

I am a newbie in digital video...
Why "All-in-one"?
Does it really work?
Smoothing: Spatial or Temporal?
Right sequence of filters?
Spots detected only at the bottom
Limit of motion table
Interlaced video?
Processing telecined video
Processing 8-mm film captures
Avisynth compatibility
Why do I get all spots highlighted and colors messed up?

I am a newbie in digital video. Where should I start from?

Start from here.

There are many other filters for temporal smoothing, brightness/contrast, pixel locking, etc.
Why did you include everything in SpotRemover?

Adding more filters to the VritualDub's filter chain decreases the overall performance.
Also, many filters work internally in YUV color space, but they have to output the result as RGB, that means an extra RGB-> YUV-> RGB for every such filter, which finally results in larger rounding errors and lower performance.
In SpotRemover, the Temporal Smoothing is integrated with spot detection, as both relate to comparing pixel's intensity in time.
Moreover, the temporal smoothing method in SpotRemover has several advantages over other filters. Color correction means simply putting different numbers to the matrix of YUV->RGB transformation. Do you think CPU will perform any slower with the other numbers?
The Pixel Locking is also more advanced than in other filters, and works on the same time buffer used for the smoothing and the spot removal.

Can I really get rid of the dust and speckles in my videos?

Yes you can, if you tune up the SpotRemover's configuration for the type and the amount of the noise in your source.
It is not possible to offer a universal set of parameters that will make any video perfect, but SpotRemover's flexible configuration can be adapted for various tasks.
If you are particularly keen on quality, use script files for the frame-based control.

What is better: 2D smoothing or temporal smoothing, or should I use both?

Think about how the human eye works.
It integrates the picture in time, while keeping the high sharpness in space.
The human eye "averages" small variations of intensity and "discards" quick spikes (i.e. spots).
And this is exactly how SpotRemover works. (See also details)
If the video is correctly smoothed in time the major amount of noise will be removed, and 2D smoothing may become unnecessary.
This does not apply to high motion scenes, where temporal smoothing is impossible. On the other hand, the necessity for smoothing of fast moving objects is also arguable, as they often have natural motion blur anyway.

What is the best place for SpotRemover in the filter sequence?

I can suggest the following order of filters (of course, not all of them are required):
1. Inverse Telecine (also see below);
2. Resize (shrinking);
3. SpotRemover;
4. 2D smoother (if you really need one).
Because Resize is a 2D operation and SpotRemover smoothes in time, it doesn't matter for the resulting quality which one is done first.
However, from my experience, shrinking should be done before applying SpotRemover because the spatial contrast (Intensity_difference / pixel_distance) becomes higher and this helps the spot detection.
Also, the processing speed will, of course, be higher for a smaller frame size.
Important note: if you aim at removing very small spots (about 1-pixel size) then, of course, SpotRemover should go before the shrinking. Otherwise, the tiny spots can get smeared after the shrinking and may pass through the spot detection routines.

Why many spots are detected in lower part of the screen but nothing in the upper one?

There is a limit for the amount of spots per frame (see Help).
Because the image is processed from bottom to top, it is possible that you reach that limit before processing the upper part of the screen.
Try lowering the spot detection sensitivity to reduce the amount of detected spots in one frame.

Why limit the amount of cells in motion table?

Each cell should have enough pixel for averaging the intensity, otherwise small intensity variation will be interpreted as motion.

Can I apply SpotRemover to the interlaced video?

Yes. There is a new option in the configuration dialog.

Can I apply SpotRemover to telecined video?
(NTSC video made from a 24fps cinema film)

This is not recommended.  (Read about telecining here.)
Instead, you should use VirtualDub's option inverse telecine to restore progressive frames at 24fps: Menu Video ® Frame rate
(VirtualDub applies Frame rate settings before feeding the video sequence to filters.)
If you still need to keep the video telecined then turn on SpotRemover's option Interlaced source and turn off Detect duplicate frames.

I captured my movie from old 8-mm film that has a lot of dust and spots.
I tried adjusting all parameters but still can't make SpotRemover see the spots.

Please see the main webpage with the news. An example of scripts solving this problem can be downloaded from the main page.
In this case, the original frame rate on the film was probably 18-22 fps, which is captured at 25/29.97 fps. This produces a video sequence with frequent duplicate frames (and flickering brightness).
This means that each spot can become duplicated in time, although originally it must have appeared on a single frame.
SpotRemover's main assumption is that a spot exists on one frame, and for each pixel in that spot there are 2 frames before and 2 frames after when that pixel is "clear". If these conditions are not met then you can't help spot detection by adjusting the configuration parameters.
Here is what you can do in this case:
1. First, it may be wise to apply the Deflicker filter (by Donald Graft), which will equalize the intensity.
Later, this will help SpotRemover to compare levels across the frames.
2. Now you need to get rid of the duplicate frames.
You can do that approximately by decimating the frame rate by a fractional number (can be done in Adobe Premiere, for example).
That number is the ratio of the captured frame rate to the one of the film.
If you don't know the latter you can find it if you take your captured video (59.94 fps) and look for a scene with clear motion.
Then by advancing frame-by-frame, you count how many different frames you have in one second. You have to look at, for example, someone walking or a moving car.
Say, within 60 frames (= 1 second) you counted only 21 different frames. Then the frame rate on the film was 21 fps.
Decimate the frame rate by 60/21.
(The speed of the video should not change. You should get the video with the same duration but fewer frames.)
3. Finally, you can apply SpotRemover.

Can I use SpotRemover with AviSynth?

No. At least for the moment.
This may be possible if you write an AviSynth script for SpotRemover.
There is a chance that sometime I will write a version of SpotRemover for AviSynth.

Why do I get all spots highlighted and all colors messed up after a minute of processing?

These are the symptoms of working in the Debug mode.
That is how the Limited length of processed video is implemented.
Unregistered version can process just enough of the video length to get an idea of how SpotRemover works and what is the function of all its configuration parameters.
The registered version of SpotRemover has no such a limitation.

More questions?

Ask me any time:

Main page

Updated: 03/05/2012