Compression Work Flow resulting in Flash

[Scott Routt]

My company has three locations. A news event is recorded with a Canon HV30 camera and a teleprompter. It is captured to a PC and exported to an mp4 (640x480) using Pinnacle. That mp4 is then FTP'd to me where I'm editing it in Final Cut.

In Final cut I create a 400x300 sequence so that I can enlarge the clip slightly. I then export as a .mov using current settings.

I convert the .mov to FLV and FTP the flash movie to my partner who puts it on a web video jukebox.

I'm not thrilled with the quality I'm getting from first transfer (Pinnacle MP4). The footage looks very flat and washed out.

The video segments are 3 minutes long and they need to be less than 100 mb for FTP.

Any suggestions?

[Tripp Woelfel]

I don't do much with Flash but I can see a couple of places where your workflow might be letting you down.

1. You're compressing the edited in a lossy format and then transcoding it to Flash. A lot of picture information is going to be lost along the way. You should output your intermediate file to a lossless format like QT/Animation for your transcoding to Flash. This may create logistical issues for you because you need to transfer the file from a distant location.

2. Your own process has caught me out. I don't understand what you're doing when you go to 400x300. You're making it smaller so that you can enlarge it? By going to 400x300 you are discarding picture information you will never get back. If you enlarge the image to fill the 400x300 frame, there's a lot of data extrapolation going on and your system is going to get it wrong about half the time. More loss there.

3. You output using "current settings"? That leads me to believe that you're recompressing to h.264 again. Loss again.

You've stepped on the footage three times in your workflow with a size 13 brougham. I'm not surprised you are less than pleased with the results.

An alternative is to stay DV through to the final transcode to Flash. Yes, DV is lossy but you'll get smaller files than QT/Animation and keep more quality than you get with your current workflow. Your process is a bit of the dog's dinner. You should rethink it and remove steps that introduce quality loss. If done correctly, your flash video should look reasonably close to your source footage.

There are also considerations on how all this stuff is shot and lit, but since you didn't give any specifics on this I cannot comment, but you might look there too.

[Seth Bloombaum]

The amazing thing is that the workflow is producing acceptable results at all. Conventional thinking, as Tripp points out, is to stay true to original acquisition formats (or improve on them using a digital intermediate), until compressing final output for distribution. Instead, you're editing a most highly compressed digital intermediate (MP4). The alternatives involve original camera tapes and fedex, or, some ftp service that costs actual money that allows gigabyte transfers.

Originally Posted by Scott Routt

I'm not thrilled with the quality I'm getting from first transfer (Pinnacle MP4). The footage looks very flat and washed out.

This is probably fixable. Pinnacle is flagging the gamma wrong, or encoding to the wrong gamma. The fact that the intermediate is MP4 is somewhat of a problem here - pinnacle isn't creating an MP4 that FCP likes.

So, your associate can do the encode with something else, Quicktime Pro to h.264 (another version of MP4) comes to mind (but I'm not sure that the gamma mismatch will be fixed - this is common with workflows that mix PC and Mac with non-standardardized file formats.)

Or, you can create a color correction preset in FCP that fixes the gamma. Depending on your hardware/software environment, this may hurt performance or require rendering.