remuxing mxf files

[deadrats]

anyone know of any (preferably free) app that can remux mxf files to something else, perhaps mp4 or mkv? i downloaded the mxf project files for Sintel, i can open the audio mxf file with every video editing program i tried but i think that may be because the actual audio is PCM, here's what media info says about the video mxf:

General
Complete name : H:\Sintel\video.mxf
Format : MXF
Format profile : OP-Atom
Format settings : Closed / Complete
File size : 24.5 GiB
Duration : 14mn 48s
Overall bit rate : 237 Mbps
Encoded date : 2011-04-14 20:20:14.612
Writing application : OpenDCP OpenDCP 0.0.0.0.0
Writing library : OpenDCP OpenDCP 0.0.0.0.0

Video
ID : 2
Format : JPEG 2000
Format_Settings_Wrapping : Frame
Duration : 14mn 48s
Bit rate : 228 Mbps
Width : 4 096 pixels
Height : 2 160 pixels
Display aspect ratio : 1.896
Frame rate : 24.000 fps
Color space : RGB
Chroma subsampling : 4:4:4
Bit depth : 12 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 1.072
Stream size : 23.5 GiB (96%)
Title : Picture Track

i wanted to run a bunch of test encodes with this really high quality source but i can't even find a media player to play it, not even vlc. i've tried googling but so far just found advertisements for some paid software that can supposedly convert mxf to other formats but in all honesty the software looks like rebranding imtoo or xilisoft software.

anyone have any experience working with 24gig mxf composed of high bitrate jpeg2000?

[poisondeathray]

ffmpeg / ffmbc can, but simply remuxing it won't help you (you will have 12bit jpeg2k in another container)

If you plan on doing testing 8bit 4:2:0 formats, then you introduce variables such as method of downconverting 12bit=>8bit and , method of colorspace transform, and method of subsampling , choice of dithering. I think sintel has lower quailty versions available that would be more ideal and standardize the method of testing

[El Heggunte]

I suppose/hope MainConcept Reference can open and convert MJP2 files with "high bit-depth". But I am not sure if all of its DirectShow filters can be used by other applications.

"Just in case", you can download the "lower-quality" pics of Sintel from http://media.xiph.org/ .

.......
anyone have any experience working with 24gig mxf composed of high bitrate jpeg2000?

I suppose the following Doom9 "elders" have it:

Mug Funky, Blue_Misfit, shon3i, mp3dom, kierank, and benwaggoner

[deadrats]

ffmpeg / ffmbc can, but simply remuxing it won't help you (you will have 12bit jpeg2k in another container)

If you plan on doing testing 8bit 4:2:0 formats, then you introduce variables such as method of downconverting 12bit=>8bit and , method of colorspace transform, and method of subsampling , choice of dithering. I think sintel has lower quailty versions available that would be more ideal and standardize the method of testing

hmmm, i did download the 4k mkv variant but i', a bit perplexed but what i see, the mxf version as media info reports, is 4096x2160 for a DAR of 1.90:1, yet the mkv 4k version is 4096x1744 for a DAR of 2.35:1; from what i have read the former is the 4k standard for film projection while the latter is 4k cinemascope, now here's my dilemma i was planning on target a dvd spec resolution with about 2mb/s, so that i show how the various encoding parameters affect final quality but i don't know whether or not i should target a 1.90:1 aspect ratio or a 2.35:1 aspect ratio.

i also think that using the mkv 4k version might skew the results, because this version is encoded using x264 at 40mb/s and in addition to testing x264 parameters i was planning on testing divx, xvid and mpeg-2, maybe even throwing in some vc-1. maybe i'm off base but i think x264 may have an unfair advantage going from x264 source to x264 target.

i'm also downloading the 4gig 720p lossless y4m for elephant's dream and big buck bunny and i'm also downloading the y4m 1080p trailer for Sintel, maybe these will prove easier to use sources.

[poisondeathray]

It doesn't matter what you use , as long as you're consistent in your testing and transparent with the testing methodology used .

ie. As long as you use a common starting point, everything should be fine. e.g. if you resize it to whatever dimension make sure you list the pertinent information like resize kernal used, whether dithering was used etc... enough info so another person can attempt to replicate your observations

The original Sintel was rendered out of blender with 16-bit RGB 4096x1744 TIFF sequence, non linear encoded . The MXF you have was prepared for projection (that's why OpenDCP was used)

If your plan was to test SD resolutions and 8bit 4:2:0 encoding , it might make more sense to start with one of the smaller resolution sequences (no reason to start with 16bit 4K )

[deadrats]

I suppose/hope MainConcept Reference can open and convert MJP2 files with "high bit-depth".

you called it, total code studio (reference has been renamed) handles them just fine.