I wonder if the metadata is not at the beginning of your video file? You might also want to use HLS streaming and segment your videos to allow easier seeking. Here's a video not suited for streaming: ftyp(qt ), Length: 12: qt , Length: 0 wide, Length: 0 mdat, Length: 20108358: moov, Length: 4689: mvhd, Length: 100: ... as you can see, the moov box comes after the mdat box, and with http, you're most likely going to be starting at the beginning of the file. Playback can't happen until the moov box has been completely processed. spatial has a --faststart flag which causes the information to be placed at the start of the file so that it's ready to play sooner.