Panasonic Gh4

Dec 122013

Recently, a comment was left asking as to why a higher bit workspace caused footage to look faded, and whether converting your footage from 8bit to 10bit or 12bit before you drop it into After Effects is beneficial. I wrote out a long reply, and figured I may as well turn it into a post!

1. Unfortunately there are so many variables involved in editing footage that I have no answer as to why your footage turns faded (could be a bug in AE, or maybe your DLSR footage didn’t like it, maybe it just appeared so on your monitor but the footage was fine?). Keep in mind, once we start working above 8-bit, our monitors become the bottlenecks; most monitors used by amateurs like myself are 8-bit, so I rely on the numeric color value displayed rather than my eyes. The best suggestion I can off is to drop a RAW image or jpeg file into the comp, and see if it also becomes distorted after edit it in a 16 bit or 32 bit workspace. If you still have problems, maybe try the same workflow in Photoshop (if you have Photoshop) to see if it’s an After Effects problem.

When I ran tests on higher bit workflows on AE CS5.5, my files did not become more faded. So its weird that you experienced that.

2. I think right now all Canons, Nikons, and Panisconis shoot 8-bit out of the box. However, you can now hack cameras to shoot RAW which is waaaaaaay better than the standard 8 bit 4:2:0 DSLR video (look up Magic Lantern RAW if you have a Canon). Many people convert their 8-bit footage to higher bit depths, but you will never get more color data from doing that. Many people trans-code their footage to  ProRes or Cineform so that they use less CPU/RAM while editing. I believe ProrRes is always 10 bit or higher, (as is Cineform, I think).

Some popular solutions for converting your video to higher bit depths are:

MPEG Streamclip, Free
5DtoRGB, (better quality, but takes longer to convert), Free
GoPro CineForm, (formally knows as NeoScene), Paid
Apple ProRes,

The main reason for working in higher bit depths is so that the effects you add to your footage benefit from more color values, not so that your actual footage gets more color values, because it won’t get more color values from converting to a higher bit depth. The colors were assigned a number 0-255, and converting to 10 bit gives you color numbers (or shade values) from 0-1023. Unfortunately however, your footage will not fill all those numbers when you convert to a higher color depth! It will have large gaps, and only fill 255 of those shades. But once you start adding effects that are beyond 8 bit, you will begin to get more color values. So the main reason for converting to 10 bit or 12 bit or 32 bit is to get more colors from the added effect layers, not from your actual source footage.

So what can you do to make 8 bit DSLR footage look like it has more color values?

One common solution is to dither footage. This is a great method of minimizing banding (banding is what you get when there are not enough colors to accurately portray a continuous color gradient).  I am no physicist, but I assume that there are infinite shades of colors, and infinite colors in our sensible world. Unfortunately our DSLRs can only capture a fraction of those colors in video mode, so when in the real world there are subtle changes in colors you notice ugly patterns in your footage because colors were omitted by the DSLR.

What dithering does is add variance to your color values by overlaying a layer of noise. This layer of noise is like film grain. And it can be so subtle, that you will not notice it. Now your footage no longer has smooth gradients. Since you dithered your footage by adding a bit of noise your smooth gradients have color variance and banding will not be as vivid because the change in color shades appears to be more continuous. It is a great way of minimizing banding. Just google “how to dither DSLR footage” or “how to dither video” to learn more.

Sorry, the comment form is closed at this time.

Film Brute