Footage with 2.0 firmware upgrade (cinematone 1 gamma)

I have been following this thread for a while now and wanted to chime in with a few things. First, in response to Kholi and the use of 5DtoRGB with Full Range selected, you are actually tossing out data when you do that. I had a long conversation with the folks that make ClipWrap, and in the end we found that what 5dtoRGB does to appear better when reviewing clips in quicktime is it compresses your full luma range in the original .mts files into the broadcast luma range in the newly created .mov files so that they appear accurate in quicktime. What it is actually doing is tossing data to achieve that compression. So while it may seem better at first, do not choose full luma and toss data, simply understand that things that appear to clip when reviewing in quicktime will actually be recoverable in the NLE when pulling down the highlights.

Next, I did a bit of testing with -15 black level, and I too am seeing negative IRE values that are fully recoverable in the NLE by lifting the blacks. Furthermore, they appear to actually be cleaner than if you had gone with +15 black levels at the start, which seem to create a tiny bit of noise and banding. I am doing all of this with cinegamma1 though as it is thus far my preferred PP gamma.
 
Whether or not the gamma has been tweaked with the new firmware, I'm really liking Kyle's PP, especially for skin tones. And with Black Levels at 0 I'm getting an image that I hardly have to touch in post. And yes, the black level essentially raises and drops the floor of the lowest black recorded, and squishes or expands the rest of the curve accordingly. However, overall things are not that simple.

I did some testing using this PP with lowering the black levels, and there is a LOT of room down in the nether lands of low IRE. I underexposed a scene and at -15 I was still able to recover some detail in the blacks. However, it was very noisy and very contrasty down there as the gamma curve would suggest, so I won't be using that setting very often. This experiment also illustrates the theory behind a higher black level. By shifting the shadows up on the gamma curve there is less contrast and more subtly in the details, at least in theory. I personally don't think it's worth the trade-off of extra noise and more difficult exposure.

One interesting thing to me about this experiment is that the histogram, which I've been using religiously in the field since I don't have any access to a waveform in the field, shows the blacks being clipped way before they actually are clipped.

I think the black level settings (and many of the other PP settings) really varies on it's overall effect to the image depending on what the other settings of the PP are (gamma, black gamma and color gamma mostly). For example, I've noticed that some settings seem to clip at 100 IRE, while others allow info above 100IRE, regardless of the knee setting. It would really be nice if Sony could publish more info on what these things actually do and how they interact, instead of having us guinea pigs figure it out.

I also find it interesting that the sensor seems to be able to record a much higher DR than we can actually access, because of the limitations of the available gamma curves. I guess this should be no surprise since it's the same sensor in the F3. But wouldn't it be great if someone could hack into these profiles and give us a gamma curve that lets us tap into some more of that range?
 
So I did a quick black levels test. I should also note that before this I tested cine1 vs cine2, but cine2 was definitely clipping black detail that cine1 was holding onto even after grading, all other camera settings equal. For the quick test I used cine1 gamma with various black levels at -15, 0, +15 both before and after grading to see what was capturing both the most dynamic range as well as detail / noise levels in the blacks. I am using the PP posted in this thread, with the exception of the knee slope set to 0 instead of -1, as I hate the plastic look that can occasionally occur with a negative slope. All the graded shots are simply graded to get as much detail in the blacks as possible with full dynamic range. Let me know what you guys think about the BL settings given these shots. I want to hear what others have to say before I chime in too much, but I will simply say 0 BL seems to have virtually 0 banding/compression artifacts.

 
Last edited:
Frank,

Do you feel like the new firmware made any sort of improvement now that you've had some time to dial in new profiles?

Cheers,
Jon
 
Postmaster said:
Yeah, I saw that - but why does he make new PPs then?
Click to expand...

My question exactly. Something was changed that brings the reds in
line from the large inaccuracies of before. Maybe it was something
else that changed this. Either way I feel it's much better after
quite a bit of checking and testing.

Just wish they would speak out about something over there at
Sony. Perhaps someone that developed some of the firmware
update or just someone in the know.
 
nemisis633 said:
Frank,

Do you feel like the new firmware made any sort of improvement now that you've had some time to dial in new profiles?

Cheers,
Jon
Click to expand...

Despite what Sony reps and Andy say, I´m 99% sure there is more change under the hood than the official list shows.
So yeah, I think - what ever they done - it´s quite an improvement. Especially regarding the reds and the highlights.

I´m ready to eat my hat in front of a camera and post that on YouTube, if the Cinegamma curves haven't changed.
 
Postmaster said:
Despite what Sony reps and Andy say, I´m 99% sure there is more change under the hood than the official list shows.
So yeah, I think - what ever they done - it´s quite an improvement. Especially regarding the reds and the highlights.

I´m ready to eat my hat in front of a camera and post that on YouTube, if the Cinegamma curves haven't changed.
Click to expand...

http://www.dvxuser.com/V6/showthread.php?281158-V1-compared-to-V2-tests

I can't tell a difference between them, personally.
 
Sorry for the delay on the Nano aVChd stills. I can't get FCP to export a still without it crushing things even further, so very annoying.
 
Kay I looked at the files from the Nano and the AVCHD, on a tripod, there's barely any difference at all between the two on a tripod as far as detailed is concerned. Hats off to the internal compression, it's quite amazing.

When moving, I've been able to pinpoint differences in detail between the two, but you'd have to really look for it (or be super anal about it) to even see it. One big thing that I've noticed AVCHD can't handle well are fast changes in exposure or exposure ramping, and this flavor is no different.

Here's the only reason why I am going to use the Nanoflash regardless of that (probably dual recording at all times), and it applies to underexposing and black gamma:

Original (+5 Gamma) - Zebras @ 85, only the reflections on the cars are clipping (click for the larger PNG screenshots):


Internal Recording (AVCHD):


Nano Flash XDcam HD422 220Mb/s:


It's pretty apparent, even in the JPGs, the Noise (FPN) and macroblocking. Which kinda doesn't matter because this is a very special case, extreme method of working that most people won't bother with.

On the XDCam nano footage, it's all fine grain/noise which is EASILY removable. Even a chroma denoise and leaving the luminance would be very pleasant.

On the topic of Black Gamma, Nirv's information has turned better results for me in the noise department and pushing extremes. Also, when combining that info with Kyle's picture profile, and adjusting Black Levels -7 versus -15, Zebras set to 95 so only the highest highlights clip on the LCD, the boosted image is frighteningly clean at 500ISO and very very heavy ND (Hoya ND8 + Cheap Fader combination, F4).

I'm still looking for a way to post some stills, but for now I can only do PNG screenshots. When I can find a good frame to post I will do so.

Thanks for all of the info in this thread, now it's a matter of balancing underexposing too far and trying to find some more detail when NDing way too much. X_x;
 
Kay, here's a quick set with a -15 adjustment on Kyle's settings. The differences in recording formats aren't nearly as obvious here, and it's still a rather extreme (purposely done) push. There's a difference, for sure, but again, not as crazy (png's on click):


Original -15 Boost (zebras showing up between the trees @ 85)


AVCHD boosted to spot noise etc:


Nano boosted to spot noise etc:


Still very impressed at how clean this camera's image is, especially when you're basically starving it of light. I can't see any of this being an issue with proper exposing before the lens.
 
High black levels don´t have to look bad:


James shot this with the following PP:

MMPP3:

Black Level: +11

Gamma Cinematone2

Black Gamma: high, +7

Knee: (too taste)

Color Mode: Cinematone2, +8

Color Level -1

Color Phase +1

Color Depth: R+4, G-3, B+2, C+1, M+1, Y-1

WB-Shift: LB-1, CC-1

Detail +4 On / Type3 Limit 7

Download original here: http://support.pro.sony.eu/software/detail.aspx#download
 
Frank,

It looks nice, but I would definitely classify it as a "stylized look" as opposed to something I would use to hand to clients for a straight out of camera shoot.

Granted, I guess that comes back to the fact that one uses different profiles for different things depending on what the job calls for =)
 
Kholi said:
It's pretty interesting... I think the "grade everything" explosion really sort of took everyone a step backward, but I too fell into it and now I'm trying to figure out if I like the muted look or the saturated and slightly crushed look. Either way, it's kinda fun.

I'm experimenting with Kyle's modified settings out to nanoFlash, recording the 220Mb/s setting because I wanted to see how far I could actually push the image down in camera (underexpose) and bring it up in post, and how working like that compared to using the internal AVCHD (which like everyone says is already REALLY freakin' good.)

Seems like the FS100 is clean enough at low ISOs to where you could do this and not suffer a major penalty, not sure though:

UnderCloseO.png
]
UnderCloseC.png

And for highlights, I let some slightly go on skin to see how it would turn out:

HighEdgeO.png

HighEdgeC.png


So far I'm pretty pleased. Still scrubbing footage, but the noise is not offensive working at extremes like this, and the color's great.

(If it's not yet realized, I am a newb to FS100 still xD)
Click to expand...


I like the look of these !
 
You must log in or register to reply here.
Back
Top