Normalize and compress dynamic range of chroma and luma

The audio editors usually have a normalizing tool that scans the whole signal to find the highest peak and normalize the signal to the highest possible level. It's posible to do that with avisynth? For example: Scan all the frames and find the most white and black parts and use that information with the levels filter. Scan all the frames and find the most saturated part and use that information with the tweak filter to saturate the to the 70% (or 85% i dont remember now). The audio editors usually have a volume maximizer (or compressor) that boost the low parts of the music to certain value and dont touch the high volume partes. One thing that usualy bothers me with some videos is that sometimes i have frames with black areas (not the whole frame). If i increase the brigness, the whle frame gets more bright. It's possible to increase the brigtness on those black spots (and only on those black spots), and leave the rest of the frame untouched. And the oposite, make the brigt parts not that bright and let the rest of the frame untouched (for viewing during night) The same with chroma, detect the parts of the frame that are less saturated and boost them a little, and let the saturated parts untouched. My goal is use that in ffdshow. thanks.

There is no single filter that will do all these things, but most of them can be done with something from this list: () with analyze and autogain parameters (sat=..., maxSat=...)

HDRAGC is probably the closest thing to what you need. it does a very nice job and is very tweakable. i do not recommend colour correction in YUV space though - it may be faster, process more efficiently and have less rounding errors and smudging from colour conversion, but it looks very unnatural unless it's done in RGB. saturation is not something to be messed with lightly... to maintain visual balance usually contrast must be increased as saturation is decreased, and that should be done in RGB space to avoid oddities and out-of-range RGB colours (like pure black luma with chroma content). as with audio editing, compression, et al should not be applied as a one-size-fits-all, but rather due care and attention should be paid to the context. colour correction is done shot-by-shot, not once for the whole film. and if something's already colour corrected and the source master is in good condition, you'll be changing the program in a way that was not intended.