HDV Pitfalls

But until I can find time, I suggest you read along with me about the inner workings of HDV, which I am being forced to embrace, because MPEG2 compression is the future of acquisition. P2. XDCAM. We can’t avoid it any longer.

1. The HDV codec is based on MPEG-2video compression, which employs both intraframe and interframe(temporal) techniques. Interframe compressors store only a fraction ofthe frames in a video as independent pictures — called key frames –and encode the remaining frames as changes relative to them. Consequently, HDV frames vary in size depending on their prior and future neighbors. In HDV 1080i, one in every 12 (25 FPS) or 15 (30 FPS) frames is a key frame. In HDV 1080p, one in every 12 (25 FPS) or 15 (24 or 30 FPS) frames is a key frame. In HDV 720p, one in every 6 (24, 25,or 30 FPS) or 12 (50 or 60 FPS) frames is a key frame.

Layterms: Each frame of action (on film) is one whole frame of action. A complete Image. Unlike standard def video, where each frame is half of the image. MPEG2 saves space by only having a whole frame of video at X amount of frames (Keyframe), and several of the surrounding frames are simply the change from the keyframe to that frame. While this saves space, it is a pain when you try to make an edit on a non keyframe. Distortion and artifacts easily occur. “This is an unavoidable characteristic of interframe compression. Since frame data affects multiple frames (and not just the one it originatedfrom), a dropout will impact all dependent neighbors. Frame-accurateediting is also made more difficult by the MPEG-2 codec. Any modifications to the video sequence require the surrounding group of frames to undergo a complete (and lossy) decompression/recompression cycle.”

2. Render times. Due to the afformentioned MPEG 2 structure, applying effects, especailly long form, can cause astronomical render times. I’ve had clients wait days while rendering out a long form piece for a DVD.

3. HDV audio uses lossy compression (MPEG-1 Layer 2) to reduce the audiobitrate to 384Kbps. DV audio uses uncompressed 16-bit PCM at 1536Kbps. I’m an audio guy, and because most of my job is restoration, this makes my job harder. Having less to work with, means my filters and tweaks are more easily noticable.

4. The standalone decks, such as Sonys M10U, only have RCA on the back. RCA RCA = teh consumer. Sorry, it’s the truth. Pro: BNC, XLR, SDI, etc.

5. Timecode. HDV employs a non standard way for timecoding, and attempting to make a dub of an HDV tape yields the timecode unusable. As if IEEE 1394 (“firewire”) wasn’t lame enough in it’s half ass, mostly not frame accurate control, we now have to throw this wrench in the works.

So, despite all of this, I need to embrace the technology and hope it improves. It may LOOK good, and better than your Best Buy Camcorder, but it wrecks havoc in post. No wonder not many broadcast facilities even support it. *sigh*.