Hacked Sony A7S?!
- Thread starter arturochu
- Start date
paulcurtis
Veteran
One of his latest tweets (that i see public) talks about Sonys abuse of the gamma curve in Slog2/Slog3. This appears to be someone that doesn't understand what a log curve actually is.
We need to know whether he is changing gamma curves or just poking at Picture Profile settings...
As for contacting him, i'm blocked from seeing his posts. I've never tweeted anything about him or spoken to him. The only thing ever written is in this forum. Seems like a preemptive block based on something to hide? I really don't know. If what he's doing is genuine and he shows that i would be a huge supporter. He must read this forum so come on Nabil, join us and explain a bit about what you're doing, what you hope to do and get us on board with you? There are an awful lot of genuine professionals here who have worked in this areas for decades, the people here are the support you need to move the project along.
cheers
Paul
We need to know whether he is changing gamma curves or just poking at Picture Profile settings...
As for contacting him, i'm blocked from seeing his posts. I've never tweeted anything about him or spoken to him. The only thing ever written is in this forum. Seems like a preemptive block based on something to hide? I really don't know. If what he's doing is genuine and he shows that i would be a huge supporter. He must read this forum so come on Nabil, join us and explain a bit about what you're doing, what you hope to do and get us on board with you? There are an awful lot of genuine professionals here who have worked in this areas for decades, the people here are the support you need to move the project along.
cheers
Paul
Last edited:
https://www.youtube.com/watch?v=BPfaIF8f5T8&feature=youtu.be
i like the way can grade N-Log and the possibility to use low Iso with good DR
Polaroid22
Well-known member
How does it compare against pp cine4? DR is probably the same?
Here is some more information on it, if there are some coders on here http://nabilfathi.tumblr.com/
Here is some more information on it, if there are some coders on here http://nabilfathi.tumblr.com/
Last edited:
paulcurtis
Veteran
He posts a CSV with 16 blocks of 64 values, so that's 1024 values in what appears to be a 1D LUT, so that's promising in terms of whether we can achieve different gammas.
The LUT isn't a log curve though, it's almost linear. (image enclosed). It's not cine either. Those NLog images can't have come from this curve, unless this is an additional curve ontop of one of the standard ones.
He appears to be suggesting he's having a bad time with haters and racists. I'm assuming he's not referring here (no idea what anyone else is saying), i don't think anyone has been either of those in this thread. Why would anyone? We're cautious about everything just because of the way this has all happened.
Omega, i guess these are coming from 8 bit XAVC files, is that right? Is there any suggestion of 10 bit files? Can you post your workflow?
You can get low ISO cine files very easily, but the log files contain more range. It appears that nlog like the cine files hold less range than log but the 1D LUT than Nabil posted covers the full input range of 0 to 1023 - so i would expect that a resulting file would hold the whole range being offered. That doesn't appear to be the case or does it now?
thanks
Paul
paulcurtis
Veteran
Okay, so 8 bit XAVC as normal. And how do you feel it differs from the normal cine curves?
Would you be able to share an original xavc file for us?
cheers
Paul
I gave it a quick test on an a7s this morning.
So far I love the way it looks, the app itself is understandably in an alpha stage and works well enough for tests but is very slow.
I also noticed my a7s getting warm which it never did previously after about 15 minutes of filming.
So far I love the way it looks, the app itself is understandably in an alpha stage and works well enough for tests but is very slow.
I also noticed my a7s getting warm which it never did previously after about 15 minutes of filming.
Samuel H
Major Contributor
It's not easy to find a scene that will show clearly where the clipping points are for both highlights and shadows. Better alternative for testing DR: shoot a shutter-speed ramp. Set your camera for proper exposure at, say, 1/400, and then shoot a video where every couple of seconds you move the shutter speed, from 1/50s to 1/3200s (i.e. from 3-stops-overexposed to 3-stops-underexposed). If the scene has enough contrast to begin with, you will be able to see at which shutter speed highlights clip and at which one shadows go useless, on each picture profile.
(if your lights are going to flicker at those high shutter speeds, you can do an aperture ramp, but then DoF can make things difficult to compare)
(if your lights are going to flicker at those high shutter speeds, you can do an aperture ramp, but then DoF can make things difficult to compare)
paulcurtis
Veteran
Thank you! At the moment there's nothing in that clip that i see that any of the cine curves wouldn't give you.
To my mind, based on the CSV file, that LUT appears to be added on top of a gamma curve which gives a slight S shape. But if you're going to add that on top of a gamma curve then you ought to do that in post. The key question in my mind is whether Nabil is able to apply a LUT to the linear sensor data and therefore create an entirely new curve based on all the data. All that existing curve is doing is raising blacks a little and rolling off the highlights. Also because of where the highlights are clipping on the other examples i don't think this is being applied to a log curve in the first place.
Ideally you want a proper log curve (there's nothing wrong with Slog2) but with lower ISOs, that would be a real achievement.
But to get there you don't want to start with a low ISO cine curve and shape that into a log curve. because you don't start off with all the dynamic range and there's nothing you'd do there that you can't do in post.
In order to see the curve in action you'd need to shoot Slog2 and NLog with the same scene. That would be a great starting place.
Mind you i haven't sold my A7s yet so i has anyone heard of any bricking? Perhaps i'll give it a go.
If anyone speaks to Nabil i would really like to have a conversation with him? PM me and i can supply emails.
cheers
Paul
Nlog vs slog
http://we.tl/exOH64gN3T
http://we.tl/exOH64gN3T
Polaroid22
Well-known member
Can you also do a NLOG vs cine4? Thanks!
paulcurtis
Veteran
Thanks for doing this!
The exposures on both of those look very different, the Slog is exposed less than half of the Nlog and the white balance is also different.
Could you let us know what the camera settings were for these two side by sides? Was the only difference ISO? I don't think so...
Stunning scenery though, whereabouts are you? Austria?
Ideally you'd want the same exposure and a scene which includes black to clipping. So if Slog2 ISO is 3200 and Nlog is, say 800 then that's two stops difference that could be compensated with the shutter speed.
Certainly the NLog has low iso and fills the bucket whereas the Slog2 version could hold at least another 2 stops of range in there, possibly more.
cheers
Paul
Nlog was ISO 100, Slog 3200, shutter 1/100, aperture 6,3, on Slog a variable ND filter (thats the problem),WB in auto, the integrated lightmeter on 0.
Thank you, i live in South Tirol, northernmost point of Italy, in the Alps.
tomorrow i repeat the shot with man.WB and compensate the exposure with shutterspeed(and a ramp).
i test also cine4
Thank you, i live in South Tirol, northernmost point of Italy, in the Alps.
tomorrow i repeat the shot with man.WB and compensate the exposure with shutterspeed(and a ramp).
i test also cine4
paulcurtis
Veteran
Lovely place to live, i was the other side of those in Austria earlier this year.
That would be great, the key is to find an exposure that shows the full range on both profiles. If you're going to be on ISO 100 on Nlog and 3200 ISO on Slog you've need to adjust 5 stops (if my math is correct) so if you're on 1/50th on Nlog then you'll need to be 1/1600th on Slog in theory. But i'm sure you know that and you'll be able to tell. You could stop down but DOF differences make it difficult to compare.
That ought to give us the same exposure to compare both approaches.
Then for cine4 it's just a case of swapping between profiles (assuming cine4 does ISO100?)
cheers
Paul
