Deinterlace/rencode of a 1080i mpeg-2 broadcast?

[mahsah]

I have recorded a video using my TV capture card in mpeg-2 TS format. If I understand this correctly, I reencode it to progressive content (thus removing the need to run a deinterlacer on playback) , essentially creating a 540p file. Are there any drawbacks to this?

Also, I am considering encoding the file with x264, but I don't know what would be an acceptable bitrate to use. I know people's perceptions differ, but most people use 700-1000 kbps for video for a DVD rip, but obviously Hi-def means higher bitrates. Could anyone recommend me a bitrate that x264 looks decent at with 1080i/540p content?

Thanks.

[LoRd_MuldeR]

The drawback on deinterlacing is that there is no "perfect" deinterlacer. If you use a deinterlacer that will keep the number of frames (KernelDeint, TomsMoComp, TDeint, Yadif with Mode-0, etc) motion might not look fluid. Using a "bobbing" deinterlacer that will double the number of frames (DGBob, Yadif with Mode-1, etc) will give fluid motion but more frames, which makes it harder to compress. Also different deinterlacers produce different "look" regarding sharpness, jaggies and combing artifacts. You will have to test a few until you find one that looks okay for you...

[Guest]

If it's a movie, it could well be telecined, in which case you want to apply IVTC, not deinterlacing.

[mahsah]

Alright. What would be a good bitrate for it, though?

[Guest]

Good is what pleases you. Find that out by trying different values.

[twolfe18]

well, i do exactly what you are trying to get into, so let me give you some advice. for me, i usually encode tv shows, so i make a half hour show (`21 mins without commercials) 350mb, and 700mb for an hour (`42 mins), which is double what you usually get on bittorrent. that works out to a bitrate of about 2200 kb/s with 128 kb/s audio (i use h264 for video and aac for audio). i keep the frame at 1024x576. i would not recommend 540p because it is not mod16, ie, 540/16=33.75 and you always want your dimensions to be divisible by 16 so that you get a whole number. this helps with compression, without mod16 dimensions, the codec has to waste bits to make the video that size.

as a general rule of thumb, you want to keep you bpp (bits per pixel) somewhere in the region of .05 and .15 for x264. the way you calculate it is: bitrate*1000/(fps*width*height). so for example, 2200*1000/(23.976*1024*576)=.156. now, some of the more advanced users will look down on this *oversimplification*, because as the video source changes, your bpp should too. things that are easy on the codec (like anime and low motion video) should have lower bpp, and things that are harder (like high motion stuff, and video taken without a tripod) should have higher bpp.

i will say though, that it isnt really realistic to go much higher than 2000 kb/s unless you just have space to fill. the is a law of diminishing returns that applies to bitrate, ie, 2000 kb/s will not look 2x better than 1000 kb/s... they actully might look kinda similar. i'm not saying there isnt a clear difference, but the differences become subtle at bitrates over 1000 kb/s.

now, the way that i chose my bitrates was by size constraints (700 mb cd), but if you dont have any size restraints, and you just want to keep stuff on your hard drive, i might recommend something like 1200 kb/s. i think that is really where the sweet spot is, IMHO. 1200 kb/s @ 768x432 would look real nice...

good luck!

[Irakli]

the is a law of diminishing returns that applies to bitrate, ie, 2000 kb/s will not look 2x better than 1000 kb/s... they actully might look kinda similar. i'm not saying there isnt a clear difference, but the differences become subtle at bitrates over 1000 kb/s.

I like your point, twolfe18
[economics lecture begins] In other words: the higher the bitrate, the less benefit (playback quality) you will receive with each extra kb/s added to bitrate

When you will equate benefit to your costs (which in our case is storage available), you get the most optimal output, where efficiency is achieved. [economics lecture finishes]