roxics
Veteran
Your link is either missing or off limits. Do you have another?
Samsung NX1 getting log-like setting
- Thread starter eddavid
- Start date
Your link is either missing or off limits. Do you have another?
Ted Ramasola
Veteran
It's for magic lantern moderators eyes only
Squig, I read the inputs from the other mods on that thread. Whats A1ex's latest input on it?
Last edited:
Samuel H
Major Contributor
If you read the whole sensor and then downsample, you get the bad RS of the full-sensor readout mode. If you line-skip you improve RS, but you can get artifacts (which can be reduced if you read the whole line and then downsample horizontally).
Apart from that, very promising. Just the fact that they gave the ML devs special treatment is a very positive sign.
roxics
Veteran
Is it raw processing power that is doing the H.265 encoding or is it a special chipset that only does the encoding? I can see your point if it's just raw CPU power, but I always assumed these cameras had special chipsets for encoding that stuff.
Huh? What are you refering to?
Samuel H
Major Contributor
Ditching H.265 will free lots of processing power, but that can't be used for a faster sensor readout since it happens in a different piece of silicon (the sensor).
In any case, the H.265 compression is probably happening in a specific hardware pipeline (one that is not capable of doing anything else).
To that link that squig posted-then-removed and we can't see because is only accessible by ML devs. I can't see what's there, but it seems like Samsung gave them some insider info.
Is it correct to assume that this software development kit is something different from the open source code that has already been released?
Samuel H
Major Contributor
Even if it's the same thing that everybody else will get access to, the fact that they know what ML is and went throught the trouble of contacting them to give them early access, or just a heads up about public availability, is still a good sign. They know who can help them, so they probably know how too, so there's a chance they'll give them what they need.
Either 2K downsampled RAW or 4K H.265 with a log profile that preserves the whole DR from the sensor would make this camera a real beast.
dxomark hasn't tested it yet, but the NX30 was already close to the NEX-5N in RAW stills, so the NX1 could be close to the a6000. Even if it's not any better than the NX30, it would still be around a stop better than the pocket, which is the video camera with the most DR for under $5K as far as I know (I tested RAW stills from my 5N against the pocket, and the 5N won by 1.5 stops, which is A LOT; pity it lost 4 stops when in video mode; the a7S has a log profile that preserves the whole DR from the sensor, so it ends up just one stop lower in video mode than in RAW stills, basically because of lost detail in the shadows because of compression; with the right profile the NX1 could be pretty close to that, but might save that extra stop too if it records uncompressed 2K downsampled RAW or 4K H.265).
http://www.dxomark.com/Cameras/Comp...amsung-NX-30-versus-Sony-NEX-5N___949_931_737
Either 2K downsampled RAW or 4K H.265 with a log profile that preserves the whole DR from the sensor would make this camera a real beast.
dxomark hasn't tested it yet, but the NX30 was already close to the NEX-5N in RAW stills, so the NX1 could be close to the a6000. Even if it's not any better than the NX30, it would still be around a stop better than the pocket, which is the video camera with the most DR for under $5K as far as I know (I tested RAW stills from my 5N against the pocket, and the 5N won by 1.5 stops, which is A LOT; pity it lost 4 stops when in video mode; the a7S has a log profile that preserves the whole DR from the sensor, so it ends up just one stop lower in video mode than in RAW stills, basically because of lost detail in the shadows because of compression; with the right profile the NX1 could be pretty close to that, but might save that extra stop too if it records uncompressed 2K downsampled RAW or 4K H.265).
http://www.dxomark.com/Cameras/Comp...amsung-NX-30-versus-Sony-NEX-5N___949_931_737
Last edited:
squig
Veteran
I'll take 2k raw over bastardised 4k any day. There's just that much more dynamic range and post manipulability. The NX1 has a stop over the 5D raw, that plus the potential for speedier and less buggy firmware updates makes it an attractive proposition. The mount has a lot of lens options too. The question I have is: is this sensor going to produce a better image than the CMV12000 sensor on the Axiom?
Samuel H
Major Contributor
* "The NX1 has a stop over the 5D raw" --> in video mode or in RAW stills?
* H.265 vs RAW: yes, uncompressed RAW will probably mean a better final image. Right now they are about the same in terms of painful workflow, but H.265 should become much easier to work with when Premiere et al start taking it as input format.
* I would probably prefer the NX1 sensor to the CMV12000 sensor, because of DR. But the CMV12000 is global shutter and can potentially run much higher fps, if the processor is up to the task.
* H.265 vs RAW: yes, uncompressed RAW will probably mean a better final image. Right now they are about the same in terms of painful workflow, but H.265 should become much easier to work with when Premiere et al start taking it as input format.
* I would probably prefer the NX1 sensor to the CMV12000 sensor, because of DR. But the CMV12000 is global shutter and can potentially run much higher fps, if the processor is up to the task.
squig
Veteran
Stills.