F3: S-Log vs Cinegamma quick look (Alister Chapman)

L

lexicon

Well-known member
Alister Chapman: "Here's a short little sequence comparing the S-Log and Cinegamma recordings from a PMW-F3. Not very scientific, both shots were exposed by setting the iris so that peak white was just below clipping. With so much white in the end part of the shot the S-log looks really overexposed until you grade it. The striking difference though is how much more detail is resolved in the dark parts of the shot with S-log showing a clear dynamic range advantage. I have also include a graded cinegamma clip where I have brought up the shadow areas to try to match the S-Log. If you look at the dark areas of the graded cinegamma you will see a lot of additional noise, a direct result of trying to lift the dark areas. This further shows the benefits of using S-Log. Interestingly the recordings were made on the SxS cards so are only 8 bit 4:2:0 XDCAM EX. The S-log graded pretty well!!"

more info and full res images here:
http://www.xdcam-user.com/2011/06/pmw-f3-s-log-and-cinegamma-quick-look/

 
Why did you choose Cinegamma 1? Did you customize any of the Picture Profile setting? And what was the color temperature set to, and can you even set the S-log color temp????
 
Thanks for posting this. It is a great comparison. BTW, for my knowledge, can one record S-Log to the SxS cards? I thought the S-Log was reserved for the dual-HD-SDI output.

Thanks
 
You can record S log to the SxS cards, or even out through the SDI output. But, you do not get the full S log curve. You only get that through the dual link. From how I understand it you might get a bit on the top and bottom, but not all the steps. I have seen it from a quick test I did and questions I have asked to Sony. There needs to be a proper test with all three types of output to see how much you lose.
 
Nigel A said:
You can record S log to the SxS cards, or even out through the SDI output. But, you do not get the full S log curve. You only get that through the dual link. From how I understand it you might get a bit on the top and bottom, but not all the steps. I have seen it from a quick test I did and questions I have asked to Sony. There needs to be a proper test with all three types of output to see how much you lose.
Click to expand...

I'd like you to explain a bit more why do you think that´s happening. You might get some posterization by grading the 8bit xdcam footage, but the dynamic range should remain the same, unless the curve is cut at the top and at the bottom. But if it were cut, then it would not be an Slog curve, isn´t it?
 
macgregor said:
I'd like you to explain a bit more why do you think that´s happening. You might get some posterization by grading the 8bit xdcam footage, but the dynamic range should remain the same, unless the curve is cut at the top and at the bottom. But if it were cut, then it would not be an Slog curve, isn´t it?
Click to expand...

I'm also looking for those conclusive answers. The answer I got from Sony was that you lost a bit at the top and bottom of the curve, when you didn't record with the dual link RGB mode, but I would like to get someone, who has a lot more technical background than I do to get into it and have a more full depth explanation. I did 3 similar shots of a small children's car with a white roof. In S log, RGB you could see more detail and gradation in the whites. The SxS and SDI outs were mostly white, without any differences in the white. I thought the same thing with the point you raised with the DR , and thought it would all be there.
 
The S-Log footage looks way better in the first part where there is a lot of mids & lows. The well lit portion actually looks about the same for both recordings. I would say that even without the 4:4:4 output it looks like quite an improvement.

Any chance this could make its way into an EX-1 via a firmware upgrade?
 
Bassman2003 said:
Any chance this could make its way into an EX-1 via a firmware upgrade?
Click to expand...

Don't know about the EX-1 but I'd sure like to see it as a free upgrade on the F3. Anyone at Sony listening?

Thanks a great deal for the test Alister. I hope the first of many we'll be seeing on this comparison. I too was wondering whether the choice of cinegamma 1 affected the results as at least in my understanding it compromises shadow detail for extended highlight range the most of the cine gammas. I suspect that cine 3 or 4 might have handled the shadows better, but perhaps there would have been corresponding loss at the top though.

I'm kind of a newcomer to using cine gammas as most of my work has had fewer extremes of contrast and I don't trust many of my clients to grade especially with flesh tone needing to be a little bit lower. However the last 2 days I was shooting with the F3 on a fast moving doc as a one man band on a chicken farm. (Actually way more fun than that sounds). Hot overhead sun, extremes of sun and shadow and people moving in and out of cages where the shadows were very dark but direct sun was still brilliant. I immediately saw the value of cine gamma over std and tended to choose 3 or 4 instead of 1 as at least on my TV Logic 5.6 " they seemed a better balance of highlight and especially shadow detail. Does that make sense to you guys?

It sure made me lust for the lattitude of Slog or an Alexa though.

Very interesting that the first test of SLog on an SxS card looks to have graded well. There is so much disagreement on the boards about this issue. Some people complain bitterly about grading it and others have no problems at all.
 
Lenilenapi said:
Very interesting that the first test of SLog on an SxS card looks to have graded well. There is so much disagreement on the boards about this issue. Some people complain bitterly about grading it and others have no problems at all.
Click to expand...

You never quite know where you want to put in a lot of work when you are shooting, sometimes DP's don't see the film again after wrap for another year before it hits color. By then their entire shooting style, choices in aesthetic have changed, and entire re lights of scenes can occur. But the worst is when you are trying to fix inherent flaws in a codec. that's more time consuming and painstaking than doing off the cuff looks. it comes right down to the amount of noise introduced in the compression itself, or trying to find a thin line in where you want the image color to float in, but there are no fine lines in 8 bit codecs. you nudge and it goes too blue. you nudge again it goes too yellow. basic contrast expansion and warmth/cool looks can be pretty easy and painless on most acquisition formats. but it's when you really want to dial in a look is where things start falling apart.

there's a reason why technicolor engineered their own gamma, log-esk curve for dslrs. it was impossible for them to match high end digital cinema cameras perfectly. or the work was so painstaking that they needed to come up with something. while they have the expertise to take any RAW or log based acquisition and match them to the point where you'd have no clue on what it was shot on. the limitations are there, it's all about what you want to do with your footage.
 
Everts said:
Is it just me or does the Slog footage that I'm seeing here look a bit soft ?
Click to expand...

It wasn't simultaneous dual recording. You can see where he's focusing in and out on both clips as he's recording, and its at different spots, so you really can't compare focus. Likewise the girls skin tones match in some places in both clips, but not in others due to the girl moving differently.
 
This looks like it's from Broadcast Asia. according to the japanese rep s-log is already out worldwide, but when i asked the same question to our local rep he said maybe 'july'. The japanese rep also answered $3,000 US for s-log.
note that the f3 there was using the SONY PL mount lens, which are not fantastic lens.
 
Everyone interested in seeing how Convergent Design Gemini recorded F3 S-Log 10bit 4:4:4 uncompressed compares to the internal 35Mb/sec 4:2:0 SxS recorded S-Log (with NO LUT applied) in a rather low light situation can download this zip...

http://ftp.datausa.com/imageshoppe/outgoing/SONY_F3/F3_SLOG444_vs_slogSxS.zip

You'll find a .dpx from the 4:4:4 RGB stream, and an 8bit PNG from the SxS. Load into Photoshop, then copy the .png over the .dpx (which expands to 16bit in Photoshop) and do a layer flick test... At least to my eyes, the dynamic range seems identical. Grading ease, compression, macro-blocking and noise structure in the various channels, well... that's a different deal and the uncompressed shines...

Regards,

Jim Arthurs


Nigel A said:
You can record S log to the SxS cards, or even out through the SDI output. But, you do not get the full S log curve. You only get that through the dual link. From how I understand it you might get a bit on the top and bottom, but not all the steps. I have seen it from a quick test I did and questions I have asked to Sony. There needs to be a proper test with all three types of output to see how much you lose.
Click to expand...
 
The 8bit will represent the slog signal, just not with the accuracy that high quality acquicition will render it.

Its important to note that while sxs will capture sLog, its benefits are suited to a higher end workflow.
 
still better than your average 709 or other gamma settings. but miles away being any sort of mastering quality. it may look like it retains all the DR, but the banding from the 8 bit compression is gonna make you lose that DR. it's really the poorest man's log recording in the world... id still probably take it over the built in 709 gamma though. really kind of despise it...
 
So, here's a question to those with some exposure to the final shipping upgrade (if any).

The early NAB era S-log upgrade I have access to does not allow white balancing on camera.

The upgrade using regular RGB 444 does, but S-Log mode does not. Switching into S-Log disables the cameras white balance switches (presets, etc) leaving the camera at 3200K.

Only the 5600CC switch has any impact, electronically changing the color temp from 3200 to 5600K. So, in effect, you have two "filmstocks". One tungsten, one daylight. All other CC manipulation is done with lens filtration.

Is this final functionality, or simply an issue with the NAB era beta? Timur, did you guys white balance on the Vegas shoot?

Regards,

Jim Arthurs
 
You must log in or register to reply here.
Back
Top