Sony FS100 AVCHD holds better in color grading than anything else

Okay, ether I´m missing something here or this is a real bummer. But from my first tests, it looks like the AVCHD codec of the Sony FS100 holds up better in color grading than anything else.


Here is what I did.

This is the original material, straight from the card, in glorious AVCHD, out of the FS100


Here is the same material – converted to Cineform (Film) and graded, to se how much detail I can recover in the blown highlights.




Here comes the same file, converted to Black Magic cocec uncompressed, 8 Bit YUV, 10 Bit YUV and 10 Bit RGB. All are graded with exactly the same settings.


10 Bit YUV

10 Bit RGB

Now here comes the interesting bit. This is the original AVCHD file, graded with the exact same settings:

I couldn’t believe what I saw. This turns a lot, of what I know about codecs upside down.
Maaybe David from Cineform want´s to chime in here – I have no idea,what´s going on.

UPDATE: For the record: I just tried Avid´s DNXHD – results are exactly the same as with Cineform or the BM codecs.


11 thoughts on “Sony FS100 AVCHD holds better in color grading than anything else

Add yours

  1. Your conversion just cuts the super white part of the source files. Find a better conversion workflow and the problems will be gone.

  2. I don’t know if I am missing something, but if AVCHD is the recording codec anything that comes out the original recorded codec is going to loose quality. It will be more noticeable where ever there is less meat to work with like on the over exposed highlights.
    If you had recorded with your BMD uncompress 422 and converted to AVCHD you would have seen a reduction in quality. Maybe not as noticeable because you started with a beefier recording format, but you will have a loss of quality.
    I guess that what I am trying to say is that when you color graded the AVCHD, you started with a beefier image to start with. The other format started from a degrade image and then it went through another degrading during color grading.

  3. I don´t know Douglas. With Cineform, maybe (it up-converts the 4:2:0 AVCHD to 4:4:4 though) since it´s not a lossle codec. but with the uncompressed BM codecs, it´s not likely that I loose quality or information.

    I gonna run a side by side test – Internal SD Card AVCHD vs. uncompressed BM and we see what we get.


  4. Looks like an encoding problem. As Jay said, you clipped your overbrights in your conversions. Try it again using ProRes.

    Apparently, the AVCHD spec calls for encoding overbrights (whites over 100 IRE or above 255 in 8-bit digital), which can be played back from Blue-ray over HDMI to SuperHD displays (10-bit+)

    So any AVCHD footage recorded to spec will have more data above pure white that can be pushed down in grading to reveal detail.

    When you convert to Cineform, although the colorspace is bigger (4:4:4 as opposed to 4:2:0) that only means more bits per register. It doesn’t mean dynamic range over 255 per register.

    From your example, it looks like Cineform is clipping your input file at pure white (100 IRE/255) so there’s no detail to recover above white.

    You’ll probably get better results if you transcode to Apple ProRes, which is a 10-bit encoding codec. It will preserve the over brights and under darks from your AVCHD, even though it can’t play them back on an 8-bit system. The data will be recoverable in grading, though.

  5. The original has the least math and the highest fidelity. That’s how ti should be. Less is always more. Transcodings will only show a quality advantage in multi-generation processing with intermediate output. When you do render the processing to a new file and start with that for the next cycle of processing. If you work on a single timeline and output from it using the original files, you can only degrade the original by converting to a different codec. And that’s when the transcoding is done properly, there is always room for errors that will degrade the output even more when using an intermediate.

    Most of the comparisons that are designed to show the advantages of intermediate formats are manipulated by an atypical workflow or unnatural processing decisions. Multiple generations that are not typical in everyday workflow or things like lowering signal level and saving to introduce aliasing, then raising level to show the “advantage” of the intermediate. The Cineform 10bit/8bit comparison was like that.

  6. I was always under the impression that converting to a “higher” codec – especially an uncompressed one – would (of course) not make the image better or ad information out of nothing, but would at least not loose information.

    It looks like I was wrong and the YUV-RGB conversions probably also have something to do with it.

    Well, that only strengthens my believe in an uncompressed workflow from recording to delivery.

  7. You messed up with the tool you used for conversion, likely something using 8-bit RGB. cgRGB has white at 255 which maps to 235 in YUV. So you lost values 236-255 from you YUV source.

    The advantage to intermediate is rarely from the first pass conversion, while there are some corner cases where it is true, the benefits of an intermediate come when you need to move an image between tools, where processing occurs between tools. Also may users now deartifact/denoise camera sources, before editing, where are you going to put those results? An intermediate or uncompressed can handle the result, whereas a camera/8-bit 4:2:0 format can’t store all the information of those enhanced images.

  8. I’ve seen problems grading a Cineform AVI vs the original AVCHD as well. I did many tests, and it’s obvious mainly in the shadows. If you need to convert to anything, try DNxHD. I noticed that it is indistinguishable from the original AVCHD when grading though somewhat larger in file size.

  9. The issue as has been previously stated is not the codecs it’s the conversion. The FS100 records “full swing” YUV Which means although in yuv colorspace, it exceeds broadcast safe levels, having a full range of 0-255. If your conversion process is just looking at values between 16 & 235 then that is the expected outcome.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Create a free website or blog at

Up ↑

%d bloggers like this: