Odyssey7Q April 2014 Firmware Update

Excuse my ignorance on this, but the press release says that the ITU709 (800%) setting "increases Dynamic Range in highlights by three stops."

Increases the DR relative to what?
l

relative to the Rec709 gamma - that's Sony's nomenclature: 200% = +1 stop, 400% = +2 stops, 800% = +3stops.

Technically the Rec 709 gamma should have only 5 stops of range, but I think they fudge the curve a bit to stretch it to 9 stops. The 800% should then give you 12 stops.
 
Has anyone used 709 800% yet? When would you use it vs S Log 2?

When you want footage that maximises the dynamic range of the camera, but don't want, or have the opportunity, to grade the footage (as you need to do with slog2). It should give you 'normal' looking uncorrected images with a bit more dynamic range at the top end (more highlight headroom).

The slog2 gamma curve is designed to maximise the recorded dynamic range of the FS700/F5/F55 cameras (13+ stops) in an efficient manner (trying to spread the important centre stops evenly across the available data points, whilst minimising the crushing at the top of the curve required to compress those 13stops in a range designed to accommodate just 5). However this gives you a very low contrast, unsaturated look that needs correction (at least a 2.2 gamma correction applied).

In comparison the ITU 709 800% profile maintains the standard ITU 709 gamma in the centre of the curve (the generally important mid-tones in the image), while applying a bit more compression at the top of the curve to get a few more extra stops in there.

You can see what's going on here:

http://pro.sony.com/bbsccms/assets/files/show/highend/img/fs700/FS700-tab1-blade5.0.png

(a true 709 curve would be straight, I believe).
 
at the first sight 4K2HD ITU 709 800% is much closer to sLog than to ITU 709.
Minimum ISO indicated is 2000 same as in sLog and saturation is also decreased.
I have to admit I used ITU709 800% setting only once when the camera was new in 2012 for
a comparison but I cant remember details.

Unfortunately it's cold and cloudy today and I could not run a test is sunny and high contrast condition.

REC 709 - highlights seem clipped, but are still recoverable

FW_APR_REC709.jpg



REC 709 - 800%

FW_APR_ITU709-800.jpg


sLog2

FW_APR_sLog2.jpg
 
Are these straight of the O7Q with no corrections? Looks like theres a lot more headroom in REC 709 - 800%...
 
Are these straight of the O7Q with no corrections? Looks like theres a lot more headroom in REC 709 - 800%...

yes, straight out of 7Q, no corrections just imported to FCP X to see the RGB parade. I have to admit that I am not sure if my profile settings are correct.

I ran another test with ITU709 800%, color level +5 to get more saturation but it seems it does not affect recording in 7Q. Can anybody confirm or correct my experience?
 
Last edited:
So long as your post workflow is correct, none of these curves are going to change the total amount of dynamic range offered by the camera. What they will do is add various curves to redistribute the range to different parts of the image - from a flatter slog log look to a more saturated 709 look. But the brightest and darkest spot in each should be the same. It's a case of personal preference as to what you want to start with.

I've done extensive testing of the range from the RAW so think that i've finally got a pretty good handle on it and once i'm 100% sure then i'll publish my results, i want to finish my due diligence before posting anything though and i'm still learning. The beauty of going via RAW is that you get all the data before any curves. You're not going to get better source data than that (AFAIK)

What will be interesting with the new firmware is whether these settings are affecting the raw data too. I think it's been said in the past that there is an effect, so i will check into this next...

cheers
Paul
 
The subtle contrasts in the highlights is what I seem to be yearning for. One has to underexpose by a good few stops to get a sky with clouds to be contrasty enough to properly see the variations in the cloud formations (which is what I really love). But of course doing this makes you underexpose the skintones.

Shooting with REC 709 helps this but I've always disliked the over saturated colors of REC 709...

Just thinking out aloud here if anyone has an opinion?
 
at the first sight 4K2HD ITU 709 800% is much closer to sLog than to ITU 709.
Minimum ISO indicated is 2000 same as in sLog and saturation is also decreased.
I have to admit I used ITU709 800% setting only once when the camera was new in 2012 for
a comparison but I cant remember details.

Unfortunately it's cold and cloudy today and I could not run a test is sunny and high contrast condition.

REC 709 - highlights seem clipped, but are still recoverable

FW_APR_REC709.jpg



REC 709 - 800%

FW_APR_ITU709-800.jpg


sLog2

FW_APR_sLog2.jpg

709 800% is more contrasty - as is to be expected. How do you mean that the highlights are recoverable? If you bring the exposure level down in FCPX does the detail in the clouds suddenly appear? That's a little unexpected.
 
The conventional wisdom seems to say that REC709(800%) is best used when in dimly lit conditions. In playing around with it, my experience is that it tends to blow out the highlights in brightly lit scenes, like outdoors. Whether the hilites are recoverable in post, as Clermond suggests, is arguable. I've found the contrast for brightly lit scenes to not be as pleasing as slog2 for the same scene. However, when used in interior lit scenes, with controlled lighting and less DR, REC709(800%) works very well, for me.
 
yes, straight out of 7Q, no corrections just imported to FCP X to see the RGB parade. I have to admit that I am not sure if my profile settings are correct.

I ran another test with ITU709 800%, color level +5 to get more saturation but it seems it does not affect recording in 7Q. Can anybody confirm or correct my experience?

I did a quick test a few weeks ago - I changed all of the PP settings one by one and monitored on the WFM on the 7Q - nothing but the ITU709 and the SLog2 modes (and now 709 800%) made any difference to the output - all other setting, from what I could tell, only affect the internal recorded image in the FS700. I guess that makes sense.
 
709 800% is more contrasty - as is to be expected. How do you mean that the highlights are recoverable? If you bring the exposure level down in FCPX does the detail in the clouds suddenly appear? That's a little unexpected.

I've generally found that by default FCPX will scale most QT to be within legal limits yet retain the superbright parts so they can be brought down, even with standard AVCHD stuff.

The question with FCPX and Slog/REC709800 is whether it understands what kind of footage it is, because under the hood FCPX is ACES and needs IDTs to bring footage in. (It may not be a full ACES environment, maybe Apple cherry picked parts of the standard). What i mean by this is does the footage react properly when you bring the exposure down, does the sky remain superbright and brings detail in or does the whole thing grey up?

cheers
Paul
 
For years a good many DPs have utilized the REC709 (800) setting as well as Panasonic's corresponding FILM REC mode to open up the dynamic range of the image without having to force the production to add a color correction step. For many productions anything but the most rudimentary post processing is just not going to happen. REC709 (800) opens up the dynamic range in the highlights, but it also means that you have no overexposure knee, and when those highlights go, they're gone. So it really is for more controlled lighting situations where you can protect those highlights from going too far, yet want to take advantage of as much of that exposure range as possible.

I, like many DPs I know, like to shoot in REC709 (800) and not do any further correction in post. We light to this gamma, use a good monitor, and what we get is what we want. Sometimes we will suggest an overall increase in saturation to bring back some color into the scene, but hat's it nd it is always a simple global correction so that post can implement it painlessly. I usually suggest an exact setting. I've been shooting this way for more than a decade.

Note that this was not the original intent of the Japanese engineers, who fully expected people to use these settings for color correction. This was pre-LOG and pre-RAW shooting days and these video cameras were just starting to be used to shoot movies. Lots of us were complaining about the baked in look and the lack of dynamic range, like shooting on reversal film, so the engineers gave us a little more room to expose.

REC709 (800) is not for everybody or every shooting situation. If you want the most choice in post, shoot RAW. Second best range is S-LOG2. But with the addition of 4K2HD shooting mode a lot of people will be using the combination of FS700 & Odyssey7Q for "better video" than the camera alone, but can't spend time in color correction. For them, REC709 (800) may be a useful option.
 
thanks Mitch, for this detailed explanation. Firmware 1.11.100 is another great improvement! Am I right that any change in PP settings won't affect 4K2HD recording? Thanks again,
 
Tested out the new firmware today for a few minutes. Not sure I see any use for the ITU709 800% option... Seems to be nearly identical to SLOG2 but slightly brighter and with highlights clipped a bit. What's the point of a REC709 option with such low saturation and that doesn't actually look anything like REC709, just less dynamic range?

Also ITU709 (not 800%) is looking a bit better but there is still some fringing and you are still stuck with the colored highlight issues where red turns yellow and blue turns cyan in highlights. Also there is a weird issue where the highlight response seems to go up to 108% and then the very brightest range actually continues back down to 104%. So the clipped whites fall around 104% but unclipped highlights go up to 108% and the very top 4% of the image highlights is a negative image. Also it's still a bit more contrasty than I'd want, with no way of tweaking that. Basically it seems like it'd be pretty similar to doing a normal HD-SDI record except that it fixes the black pixel issue (but still with some colored fringing) but at the loss of any image control as all picture profile options are bypassed. I imagine some sharpness may be gained by using CD's debayer/downscale as well, but I didn't really test that.

I'd really love a mode that just records a nice image and avoids all the color and edge issues the FS700 has... like just set the camera to SLOG2 and be able to monitor and even record with Sony's SLOG2-to-REC709TypeA LUT which seems to do a pretty good job. The LUT that is currently in there for SLOG2 seems pretty useless to me since it is very hard to imagine ever wanting to see what SLOG2 looks like with 3 stops of highlights clipped and the whole DR stretched way out revealing a lot of noise without even normalizing the shadows. If I were recording in SLOG2 I'd want to see some kind of LUT that might be the LUT the colorist could start with or the editor or dailies might use. The LUT in there is not that LUT. Would it be hard to just switch out that terrible looking LUT for the one Sony provides for SLOG2 to REC709 Type A (seems to be the best one).

A couple times when using the ZOOM function (on two separate occasions after reboots) on engaging 1:1 or 2:1 zoom the furthest right column of the video signal would be copied across the right half of the screen. The issue was solved with a reboot. Also once when using the playback function, there seemed to be an issue where the right side of the image wrapped to the left side of the screen for maybe about 100pixels.

I haven't bought the FS700 addon for my 7Q yet and havent taken the 7Q on many jobs since it still feels very unfinished and limited as a recorder. It would be tough to go backwards in terms of playback and monitoring. The playback function that is in there seems to be designed for tradeshows or looping playback for displaying test footage, not for actual production. Any kind of simple playhead controls would really go a long way, not to mention eventually thumbnails and touching to drag the playhead or maybe dragging to jog/shuttle...

Those three images posted above don't seem to match my experience, perhaps exposure was inadvertently changed in between gamma changes? Or perhaps SLOG2 and ITU709 800% images are switched? I have found at 0db ITU709 800% is just a brighter, slightly clipped version of SLOG2, but the images above show the opposite for some reason. Also in my quick test, ITU709 seems to avoid clipping highlights better compared to ITU709 800% at the same gain setting, which seems quite different from Clermond's findings.
 
Last edited:
Back
Top