I’ve been using Shotcut to convert DV video (the 16:9 variant of DV) (854 x 480 interlaced at 29.97 fps) to h.264 / MP4 (NTSC 854 x 480 progressive at 29.97 fps). The main reason I’m doing this is because DV video is about 4 times larger than MP4 video (250 MB of MP4 video = about 1 GB of DV video). But I’m also doing it to convert the video from interlaced to progressive, and MP4 is a more widely-used / compatible format than DV. But I don’t know if I’m doing it correctly.
I’ve read that converting interlaced video into progressive video in Shotcut will cause half of the temporal information of the video to be lost. Is this true? Will half of the temporal information get lost if 480i at 29.97 fps is converted into 480p at 29.97 fps?
I’ve also read that people like upconverting 480i video to 720p or higher, and maybe at a higher fps like 60 fps – I want to do this, but first I want to create a file that more closely resembles the original (the same resolution and frames per second, but in MP4 format instead of DV, and progressive instead of interlaced).
Exporting the DV video at CRF 17 may be a good enough CRF to preserve quality, but I’m not sure.
I’ve noticed that an old version of Shotcut (the December 2019 version) appears to handle DV interlacing correctly, but newer versions of Shotcut (the January 2022 version and the April 2022 version) do NOT appear to handle DV interlacing correctly. In the 2022 versions, the DV video has a weird “doubling” effect, in which the video is visible twice and one of them is slightly farther ahead in time than the other one, causing the video to be blurry. This does not happen in the December 2019 version. Has support for DV video been dropped in the new versions? And is this an interlacing problem?
Is putting an interlaced DV file into Shotcut (and then exporting it as a progressive MP4 file) the correct way to convert DV video to h.264 / MP4? Will it decrease quality? Does deinterlacing need to be done outside of Shotcut (if both the source and new file are 29.97 fps)? Should I first use some kind of deinterlacer in HandBrake (like Yadif and Bob) before putting the video into Shotcut?
I’m also converting the video from 601 colorspace to 709 colorspace, which (I assume) is the correct way to modernize the video.
I reproduce that. YADIF broke when we introduced support for time remapping.
To do what you want by increasing the frame rate within Shotcut requires 2 steps: Properties > Convert with the Deinterlace option, and then another export to reduce the bitrate. Alternatively, you can step 1, go to the end of View > Application Log, get the ffmpeg command line, adjust it for your desired GOP and crf, and then run ffmpeg manually.
In an older Shotcut version (such as the December 2019 version), if an interlaced video at 29.97 fps is converted into a progressive video at 29.97 fps (the frame rate stays the same), then will half of the temporal information in the video be lost, and will there be a reduction in quality?
(Conversion in the old Shotcut would be by dragging the interlaced DV video (at 29.97 fps) into the timeline and exporting it as a progressive MP4 at 29.97 fps)
Is using the “convert” option in “properties” to convert an interlaced video to progressive (in a new Shotcut version) better than exporting an interlaced video to progressive MP4 (via the timeline) in an old Shotcut version? (My guess is that they would both be the same, but that the newer Shotcut would require an extra step (converting to edit-friendly)).
Regarding Yadif not working correctly in later versions: What is the first version in which it didn’t work correctly? (For example, did Yadif / interlacing work in the March 2021 version?)
Also, is the “854 x 480 16:9 NTSC” video mode the correct mode for 16:9 standard definition video (which has a default resolution of 720 x 480, but is in the 16:9 aspect ratio instead of 4:3), and will that affect interlacing?