Experiments in SLog

paulcurtis said:
Sam is along the right lines and maybe we should make Sam the designated camera tester!
Click to expand...

Unfortunately Ive spent a long time playing with profiles/looks/post fixes on cameras, and come up with my own theories on what make a camera look nice, which is more colour fidelity/tonal graduations/blocking/banding than absolute DR.

The normal tests dont hone in on colour fidelity, and graduations/blocking/banding

In 2005 the Sinarback (the first non tethered 5k,16bit,camera ever - which was so basic, being pre DNG, that it had a colour file and a black reference file for each image) I owned was a challenge to say the least, the Sony FS100 is/was tricky too.

In my digital journey the nikon D3 (in stills mode) and the 5d2 (in video mode) have been exceptional in that I performed no optimisation on them in any manner - they just looked really good out of the box

(of course the 5d only looked good at about 480px, and in flatish light, but is still looks 'good' the colour 'tune' is pleasant)

To some extent that is the test - does it look good.

And to some extent a log camera, without properly (manufacturer supplied) created LUT and delog curves for post is in some manner putting too much weight on the user.

And that is where raw/linear comes to the fore.

The user is presented with a handleable post set of options.

A raw file looks good out of the camera (having been 'auto-delogged') the user is then presented with simple exposure comp slider, or can dig into their raw software to go in as deep as they want in terms of more complex tweeks

Delogging images and having to expose for weird mid points (like 32%) is really a little to complex for shooters who want to shoot.

S
 
Last edited:
Yes, I understand what you're saying Paul, it's just that 2 stops of a difference that I'm seeing between the two cameras in these two separate tests is a big difference and maybe such things as looking into different channels or combing channels into RGB or looking only at luma in a composite signal shouldn't produce such a vast difference. And adding the F3s bigger pixels into the account I'm sure the F3 has a wider DR if not by 2 stops then at least by one and I'm sure you'd agree with me on this?

Anyway, it looks like somebody will have to get this cameras in their hands and make a test for us if not I won't be able to sleep:) Just kidding, but I have been on the fence for buying a new camera for the last month and this DR debate is the only thing that's keeping me from pulling the trigger on the F3 or the FS700... I mean if the F3 has a 2 stop advantage I definitely prefer it over the FS700 (doing docs need as much DR as I can get).
 
KINOKS said:
Art Adams charts are not done with the Slog2 however...
Click to expand...

morgan_moore said:
I thought the Zacuto tests were 'useless' at comparing cameras - as the DPs could add more fill light. What they proved is that given enough lights you can make a narrow DR camera look good.

In the real world we rarely have enough lights!
Click to expand...

I actually think that's what makes the Zacuto shootout the best comparison: you get the same scene and DPs work to tweak it to get the best result possible for each camera. That way you get to compare each camera 'at it's best' in a real world (drama world) scenario - when you see the results side by side, you can easily determine which camera produces the best image.

If you're talking documentary it's even easier, as you just need a basic stress test eg unlit person in front of a window.
 
Well the Zac test has its place - but remember they were just chucking jokers galore in to fill like they were dedos, not in my budget :)

Yep the window shot - that is what I have done with the F5, but better.. get someone outside the window wearing a flouro bulders jacket - nice.

S
 
KINOKS said:
I mean if the F3 has a 2 stop advantage I definitely prefer it over the FS700 (doing docs need as much DR as I can get).
Click to expand...

I've shot two docs recently with the FS700, making use of slog2 with both - although I haven't put it side by side with an F3, but have put it against a C300 which has 12 stops DR (in cinelog). The Fs700 with slog2 had more. We were shooting on a rooftop in bright sunshine, subject in shade. The FS held the clouds and the subject. The C300 lost the clouds. The shadows were noisier on the FS700, but they were useable, in my opinion, so I decided I'd avoid clipping and take the noise. As Paul has repeatedly pointed out you from a 4K source you have more resolution to perform noise reduction on to clean up the shadows for an HD. Alternatively you can over expose, lose a couple of stops of DR and the noise. Either way, I personally prefer to have the option.


Now, given that you need an external recorder to shoot Slog with the F3 anyway, I'd say the obvious choice is the FS700 as you get the same DR (or more), plus the option of 4K, 240fps 2K, 10 bit 444 (eventually) and so on. It's also lighter (albeit perhaps with poorer build quality). The only downside is that, if you need to operate without a recorder, the F3 has better internal compression.
 
Collin thx, that's good info! If you saw more DR in the FS700 than in the C300 that would put it around 13 stops. Regarding the FS700 vs F3: F3 doesn't need a big 7.7" recorder to shoot 10bit. The 7Q is a great option but for doing docs it's really big so I'm kinda sceptical about it. And shooting 4k, well, I think not:) To much data for the kind of docs that we're doing (verite).

So not to derail this topic to much, you said you subjectively saw more DR in the FS700 than in the C300. That kinda negates our findings in this topic...?
 
KINOKS said:
Collin thx, that's good info! If you saw more DR in the FS700 than in the C300 that would put it around 13 stops. Regarding the FS700 vs F3: F3 doesn't need a big 7.7" recorder to shoot 10bit. The 7Q is a great option but for doing docs it's really big so I'm kinda sceptical about it. And shooting 4k, well, I think not:) To much data for the kind of docs that we're doing (verite).

So not to derail this topic to much, you said you subjectively saw more DR in the FS700 than in the C300. That kinda negates our findings in this topic...?
Click to expand...

Yes and no. As I've been trying to point out, the DR depends a lot on your tolerance for noise in the shadows. In my experience I'd say 13-14 stops, it was appreciably more than the C300 - but a lot of that range relied upon very noisy shadow stops: you could see the broad tones, but no detail amongst the noise. In a documentary situation I'd happily accept that, for commercials and drama I'd want cleaner blacks and that means choking the image more - which puts it at 12 stops, with the last one being a little noisy (as it is in the C300 and F3 in my opinion).

As regards the difference between the F3 and FS700 being used for documentary: sure, you don't need a big unit like the O7Q for the F3, but you need something and the F3 is heavier and bulkier to start with. So once you add a hyper deck shuttle, the difference in terms of weight, bulk and cost between an F3 package and an FS700 package will be minimal. But with the FS700 you'll have 12 bit RGB 444 compressed (eventually) which will be higher quality than the F3 kicks out, at a very manageable data rate, plus 4K and 240fps should you need it - and let's not forget: a decent 7inch monitor too!

Having said all that... Usability might still be an issue. The HDMI lag on the FS700 might be a problem for you. The non-broadcast internal (backup!) recording might be a problem. The noise in the extra stops might be a problem. The relatively fragile nature of the FS700 might be a problem.

Honestly, if you want my advice, I'd recommend heading over to your local Sony Broadcast dealer/rental house and try them both out. Shoot some side by side comparisons, check out the footage, see how you find them to work with. Better yet, hire each of them for a job - see how they perform in the field. This sort of testing will be far more valuable to you than reading a whole bunch of contradictory posts on a forum!
 
Right, in the spirit of double checking i managed to do another test using a reflectance chart and also being more careful to eek every little bit out.

http://blog.inventome.com/Blog/2013/10/fs700slogredux/SLog2-Redux

AFAIK the ND filters on the 700 are 2 stops, 4 stops and 6 stops of ND. I assume these are accurate, if not then the results will be incorrect.

I mentioned in the article that i should be able to side by side with a BMD camera soon, although i will probably wait til i get the O7 to compare like for like really. I've also been writing up my notes on LUTs, grading and linear space. I'm hoping to get some real world examples of workflow and the pitfalls. But that's a background project.

cheers
Paul
 
Thx for another great test Paul. So results are basically the same. The 12th stop is "gravy" (as Art Adams like to call it). Usable range: 10.5 - 11 stops. Right?
 
KINOKS said:
Thx for another great test Paul. So results are basically the same. The 12th stop is "gravy" (as Art Adams like to call it). Usable range: 10.5 - 11 stops. Right?
Click to expand...

I think this one shows a little more. It displays the results differently, the step wedge is a waveform which is more black and white (forgive the pun), whereas actual images you can decide for yourself at what point you'd stop counting. In this particular case i would say there are very clear 12 stops (from same chip to chip) plus an additional stop or so visible as well, although perhaps not that useable.

The key is whether the extreme tops and bottoms of the scale would become more useful with RAW/Linear output. There's a chance that may well be the case. So i'd say that the 700 sensor can physically see 13/13.5 stops but of course depending on how you choose to test that.

As i mentioned i'll be able to get my hands on a black magic in the next few weeks and perhaps the O7 recorder will be around then too - in which case that would make a nice comparison i think. It would be side by side comparisons in the same conditions that would be more useful.

cheers
Paul
 
The wedge values that fall below 10 IRE all fall into a zone referred to as black with no detail in the Adams zone system. Visually they don't amount to much.
 
Razz16mm said:
The wedge values that fall below 10 IRE all fall into a zone referred to as black with no detail in the Adams zone system. Visually they don't amount to much.
Click to expand...

Yes. These just contribute to a greater perceived dynamic range.
 
Could someone explain the math formula how S-Log or a log mapping works ?
So, I have a 12-bit RAW file. We are looking at 12-stops, 2^1, 2^2....2^12

Now, I want to map this to 10-bit, so I have 2^1,2^2...2^10
I can divide the shadows to keep details.
2^1, 2^2,...2^7,2^8 = 0-255

Now I have 2^8 to 2^10 = 1024 - 256 = 768 mappings
And I have to map 2^8 to 2^12 = 4096 - 256 = 3840

768 mappings can be divided into 4 regions - of values 192 (896/4) each

Now I can take 2^8 - 2^9 and map it to 256 to 256+192 = 448
2^9 - 2^10 goes to 448 to 640

And so on...

Is this how a log mapping would work ?
 
I dont really understand your question. The mapping is defined by its gamma curve. It has nothing to do wether the log is 8 10 or 12 bit, the values are different but the factor stays the same. U cant really put it in math because there is no linear mapping, its a logarithmic mapping, like its name :D
Here are the slog1 and slog2 values for 10bit
SLog1_Slog2Chart.jpg
 
You must log in or register to reply here.
Back
Top