Removing the noise only in a part of the frame
Posted: Sat Feb 11, 2006 2:26 pm
Yesterday I was testing how NV handles clips that origin from VHS tapes. The results were overall good. Calibration was a bit tricky, but I return back to this subject later on.
Because I have (so far) only the demo version (VirtualDub plugin), NV removed the noise from the top 2/3 of the 720*576 frame (Yes, I live in Europe!), since the promised size 640x480 was exceeded. This was neat, since 1) I was able to see the results in spite of the fact the demo limitation was exceeded, and 2) I could easily see before/after quality.
This leads directly to the following suggestion: Please add an option to NV that enables to specify that only a part of the frame is affected. There are two advantages: 1) easy before/after comparison and 2) faster rendering in the testing phase.
After implementing this feature, making before/after posts to this forum is very simple: just include a single partially rendered clip. You'll save disk space in storing a single clip instead of two. The rendering time in NV is really an issue, and every attempt to decrease it is in my mind justified. If you render, e.g., a quarter of the frame in the testing phase, your processing speed will be considerable faster. In practice several tests must be made before the final rendering. It is necessary to see how to settings affect to the video, seeing a single frame is just not enough. For example, the rendered preview window seems to be ok, but playing the video reveals that the unfiltered low frequency noise will make the video worse that the original was.
How to implement this? I suggest two ways: 1) the user selects from an additional dialog half frame/quarter frame and the right/left/top/bottom, or 2) the preview window in noise filter settings is reused to specify the region that will be filtered (when the preview window is active, select partial render from the tools dialog).
You could go also further with this: the (non overlapping) frames in variant selector are optionally rendered simutaneously with respective settings. This would enable the user to see a real side-by-side comparison how different settings affect ....
Because I have (so far) only the demo version (VirtualDub plugin), NV removed the noise from the top 2/3 of the 720*576 frame (Yes, I live in Europe!), since the promised size 640x480 was exceeded. This was neat, since 1) I was able to see the results in spite of the fact the demo limitation was exceeded, and 2) I could easily see before/after quality.
This leads directly to the following suggestion: Please add an option to NV that enables to specify that only a part of the frame is affected. There are two advantages: 1) easy before/after comparison and 2) faster rendering in the testing phase.
After implementing this feature, making before/after posts to this forum is very simple: just include a single partially rendered clip. You'll save disk space in storing a single clip instead of two. The rendering time in NV is really an issue, and every attempt to decrease it is in my mind justified. If you render, e.g., a quarter of the frame in the testing phase, your processing speed will be considerable faster. In practice several tests must be made before the final rendering. It is necessary to see how to settings affect to the video, seeing a single frame is just not enough. For example, the rendered preview window seems to be ok, but playing the video reveals that the unfiltered low frequency noise will make the video worse that the original was.
How to implement this? I suggest two ways: 1) the user selects from an additional dialog half frame/quarter frame and the right/left/top/bottom, or 2) the preview window in noise filter settings is reused to specify the region that will be filtered (when the preview window is active, select partial render from the tools dialog).
You could go also further with this: the (non overlapping) frames in variant selector are optionally rendered simutaneously with respective settings. This would enable the user to see a real side-by-side comparison how different settings affect ....