Slog

M

Mestizo Devon

Well-known member
Is it possible for the FS100 to run SLOG?
Not that sony would do this...
The C300 has 8bit and an modest chip (11K +/-)
Then the F3 factor
Yet I wonder, any thoughts???
 
No, not really.... there's some "logish" profiles, IMHO they are a clear improvement over the default picture profile, but are not even close to what SLog/444 from a F3 will give you to grade with.

We can get our FS pretty close to an F3, when the F3 is in 709.

But we cannot get the FS even into the same playing field as Slog/444.. but for that matter matching a F3 in 709 to an F3 in Slog means dumbing down the Slog camera, as the 709 will never rise to match the Slog, and either will a FS no matter what profiles are in the camera.. unless Sony releases an update to support it.



d
 
Last edited:
S-log on F3 is only possible via dual link SDI out. It's safe to say S-Log is not possible with FS100.
 
Mestizo Devon said:
I've used franks profiles and like them a lot
I'm wondering why is it difficult to add to firmware
Click to expand...

It is not a question of firmware - S-Log is a feature Sony wants to sell for $$. If they put it in the FS100, many customers would buy the cheaper camera and Sony would lose F3 business. They are protecting the business viability of the entire camera line. Some news today....the F3 S-LOG option had a dramatic price decrease and you can get S- LOG 4:2:2 now.
 
rx8pilot said:
It is not a question of firmware - S-Log is a feature Sony wants to sell for $$. If they put it in the FS100, many customers would buy the cheaper camera and Sony would lose F3 business. They are protecting the business viability of the entire camera line. Some news today....the F3 S-LOG option had a dramatic price decrease and you can get S- LOG 4:2:2 now.
Click to expand...

It's impossible to implement something as S/Log without an uncompressed out right from the sensor. It's a hardware issue, not firmware or some devilish corporate plan. It's like saying your FIAT doesn't do 250 MPH because they are protecting their Ferrari line... :)
 
jetswing said:
S-log on F3 is only possible via dual link SDI out.
Click to expand...
\\

That is incorrect. S-LOG can be recorded on any external recorder (single or dual-lin, or even HDMI), or as someone else already said, even to the internal SxS cards.
Whether that is a good idea or not is another story, but technically S-LOG can be recorded anywhere.
 
taubkin said:
It's impossible to implement something as S/Log without an uncompressed out right from the sensor. It's a hardware issue, not firmware or some devilish corporate plan. It's like saying your FIAT doesn't do 250 MPH because they are protecting their Ferrari line... :)
Click to expand...

LOG encoding is merely a gamma curve that places the visually more significant bits of data into a small container to give the user more flexibility in post at the expense of making an unusable image out of the camera . It has no relationship to compression at all. The internal DSP of the FS100 is done before compression which includes the gamma. It is not to be confused with RAW coming from a masked sensor which is totally different. There is no hardware reason at all that the FS100 could not make a log curve but it will not output 10bits like the F3 which would keep it away from that performance territory.
 
Makes sense...

i graded a feature shot on F3/Slog/444/Gemni/ProRez, and some splinter unit shot on F3/709/internal codec & one pickup scene shot on Alexa/444/ProRez

Aside from the scaleing hit taking 1920x1080 to 2048x1080 from everything, the only issue was making the inserts shot @ 709 in a F3 to match the Slog F3, and that was a hill that was not going to be climbed easly... trust me on that one, i have the been there, done that t-shirt now.

But the compression from either the Sony onboard Codec or the ProRez was not the issue in matching the 709/F3 to Slog/F3.. it was the increadable range of the Slog & Log-C camera's -VS- 709.

d
 
rx8pilot said:
LOG encoding is merely a gamma curve that places the visually more significant bits of data into a small container to give the user more flexibility in post at the expense of making an unusable image out of the camera . It has no relationship to compression at all. The internal DSP of the FS100 is done before compression which includes the gamma. It is not to be confused with RAW coming from a masked sensor which is totally different. There is no hardware reason at all that the FS100 could not make a log curve but it will not output 10bits like the F3 which would keep it away from that performance territory.
Click to expand...

Thanks for the correction, you got the technicalities right while I got them wrong, point being, there is no need to protect the F3 via firmware crippling, since S-LOG on the F3 only exists because of the dual link SDI that outputs 10-bit. The F3 firmware may not get an uncompressed image, because there is A/D first, but gets an "unlinearized" log image, the way it was exposed in the sensor, before it has been converted to linear color space, that's the point.
 
taubkin said:
Thanks for the correction, you got the technicalities right while I got them wrong, point being, there is no need to protect the F3 via firmware crippling, since S-LOG on the F3 only exists because of the dual link SDI that outputs 10-bit. The F3 firmware may not get an uncompressed image, because there is A/D first, but gets an "unlinearized" log image, the way it was exposed in the sensor, before it has been converted to linear color space, that's the point.
Click to expand...

Couple of things....just because this can very quickly become confusing.

The beginning of the process is the analog to digital conversion. At this point the processing can begin. The native sensitivity curve of the sensor is not usable for just about any application and must have some sort of gamma correction - LOG or Linear. At the front end of the camera you have TONS of information available compared to what actually gets recorded or output. I am not sure about the FS100 but it is likely processing 12bit 4:4:4 at the DSP (yes I am guessing to some extent). This is where the gamma curve is best applied to prep all of the image data for its eventual beating as it gets truncated down to 8bit 4:2:2 or 4:2:0. In 12bits you have 4096 levels of per color channel - by the time it gets to the end you only have 256 levels per color. It is easy to see that huge amounts of data get thrown out the window never to be seen again regardless of your recorder, NLE, monitor etc. The typical 709 gamma curve is applied to push the image into a pleasant looking image at the expense of killing off highlight and shadow detail. The LOG curve is not the native curve of the sensor but rather a curve applied to stuff the most significant bits into whatever output is available. It is not a good looking picture because it looks flat and generally ugly, but it has preserved as much of the highlight and shadow detail as possible for correction that will be required. LOG gamma does not care if the final output is 4:2:2 or 4:4:4 or 8bits or 10bits. The purpose is to make the best use of limited information available. In the case of just about all the cameras - HUGE amounts of information enter the lens and a TINY amount actually gets captured and recorded. LOG Gamma helps us make the most of our limitations as far as dynamic range is concerned.
I hope this was helpful....
 
This is very helpful. I understand what you are saying, I 'm not familiar all of those numbers like you are, but I'm familiar with working with those cameras.

From what I understand, and have seen, there is a question about information vs. useable information. To some extent, you can say you have information up to seven stops of underexposure in some cameras, but that is totally ruined by the amount of noise and compression artifacts. By applying a log gamma on an image that is not that clean, you end up compromising important parts of your exposure range in favor of others (like using Technicolor profile on the 5D). So you have more highlight exposure at the cost of more shadow and mid noise.

Recording 10 bit out of an F3, you can see S-log as a win-win situation, because you gain exposure latitude, without loosing a lot of shadow and mid cleanliness. If you record an S-Log curve out of the FS-100, you may get something similar to that technicolor profile on the 5D. There isn't any possible way, that a s-log curve on the FS-100 can look as good as an S-log curve on an F3+external recorder, or give us 13+ stops of latitude, even though they share the same sensor, because the massive gain in latitude comes from the fact that you can linearize the curve later in post with very aggressive correction, not true with the FS-100.

Meaning that independently if an S-LOG firmware could be potentially beneficial to FS-100 users, it can be expected to look a lot worse than the F3. You cannot hope to underexpose 4 stops and get your exposure back on the FS100 8-bit image, like you do when shooting s-log on the F3.

If I am wrong in assuming so, please correct me!

cheers
 
It could be possible for sony to add a log style output, similar to the C300, an 8 bit log perhaps. Remember that the C300 really is just 8 bit but the log output i've seen seems really workable.

I think the FS suffers from too many picture profile options, they could be simplified quite a bit.

However we don't really know whether there is enough processing horsepower in this little box to do it. It seems that the debayering is quite different to the F3, i'm noticing all sorts of artefacts that i've not seen on F3 footage (i can't do side by side though). Also some of the noise handling and fine details in low light seem a little different too.

I see the slog upgrade for the F3 has been drastically cut in price.

But back to the point - i would love for sony to offer a more raw like log output. I believe they could. I assume the 8 bit HDMI restriction is bandwidth based but making the data really truly log would help.

I wonder how hackable the firmware is :) I think it's running a linux core (but that could be misinformation i've read elsewhere)

cheers
paul
 
You must log in or register to reply here.
Back
Top