Rendering size: 1920x1080 => 1920x1088

Category:support request
Project wiki:Blender wiki

In the blender render / Dimensions property panel, I have set the resolution to be full HD: 1920x1080, 100%, etc.

However, the output video is 1920x1088. I can verify this when playing the clip with mplayer. Also, when I import the clip into a kdenlive project (set as 1920x1080), it warns me that the clip's resolution (1920x1088) does not correspond to the resolution of the kdenlive project (1920x1080).

How to explain this discrepancy and how to solve it?


Either render to a PNG image sequence (recommended as it is much smaller) or to raw AVI (can make huge files!) from Blender.
Echoing (more or less) Chipmasque's suggestion, I would advise dealing with the imagery first, and the video playback files second, "and never the twain shall meet."

Generate your animation frame-by-frame to MultiLayer OpenEXR output files, frame by frame, right up to and including "the final print." (Which, if not actually MultiLayer, is still OpenEXR.)

Now... the process of generating the (MPEG AVI | QuickTime | what_have_you ) output files is an entirely separate and now purely-mechanical (non-)issue. At this point the process consists purely of encoding, and each process of "generating an output file in what_have_you format" is the same: it consists of taking the same, very large and very detailed and basically un-compressed but also loss-less source data, and encoding it into a particular format. This is a purely output-only process: the resulting files will not be used as a subsequent input to anything (except possibly "video editing").

I would generate to OpenEXR, even though it is sometimes huge, because basically "my hard drives are huge-er." I want to do the whole thing "a frame at a time" (so that I can always start over without, y'know, starting over...), and I want to defer the whole process of generating "a video output file" as a separate and disjoint operation. As an afterthought. I want "video output" to be the first, and the only, point at which I consciously "throw any data away" (knowing, of course, that absolutely none of it has actually been lost).

(I strongly prefer that the workflow ultimately contains only one actual digital-compression step ... other than of course lossless "run-length encoding" ... and that this is only in the "afterthought," where the real work is done, and so it actually sacrifices nothing.)
I use OpenEXR specifically to avoid any "double compression." I don't want the compression algorithms, say, of PNG, to "digitally interfere" in any way with those of, say, MMPG. (I've seen what I guess would be the digital equivalents of Moire patterns...)