Sony NEX 5N hack is go - 46Mbit 1080p tested and working

[Andrew Reid]

[img]http://www.eoshd.com/wp-content/uploads/2012/08/nex-5n-canon-samyang-and-sd.jpg[/img]

[url="http://www.eoshd.com/comments/forum/12-sony-nex-hack/"]I have set up a dedicated forum for the Sony NEX Hack here[/url]

A guy named Someone 1.00 is able to hack the NEX series of Sony mirrorless cameras for higher bitrates in video mode and more. [url="http://www.eoshd.com/comments/forum/12-sony-nex-hack/"]Discuss the hack here[/url] and support Someone 1.00 in his efforts.

The video after the break shows the hack in action, and proof that it can be done.

[Caleb Genheimer]

I think this is yet another sign of how things will begging in the camera business from now on. Until such a time when there is a full-featured, high quality cinema camera for under $4K, enthusiastic cinematographers will continue to attempt and make their own. The GH2 is a prime example of this, along with the NEX body kit as well s ReWo and other high quality cages and gear. This goes to show that many are still looking for much more than what is currently available from any camera out of the box. The first camera company to realize this and cater to it would be able to capture a large and loyal user base very quickly if they played their cards right.

The latest offerings are much closer but still leave something to be desired. The BMCC has a smaller than S35 sensor, as well as completely unnecessarily long FFD. The Sony FS700 does not yet shoot 4K, and does not deliver RAW, but is otherwise incredibly well-endowed. Canon is either using old sensor tech, or slaps on a $15K price tag. Also, rolling shutter is still an issue basically across the board.

If I were a betting man, I'd put my money on Sony coming up with the first truly great RAW camera. The FS cams are a perfect candidate, and the BMCC is almost certainly forcing many of the big players to seriously consider RAW offerings. They have all the trappings of a serious video camera (ND, XLR, etc.), as well as the incredibly short FFD NEX mount and a S35 sensor.

I have a sneaking suspicion that there is another race happening in the shadows, though . . . one between RAW cameras and compressed codecs that are very nearly just as good. Compressed codecs have the potential to be every bit as good as shooting RAW once sensor DR reaches a certain point, assuming the DP is careful and exposes correctly. It may take until after several good RAW cameras are on the market before folks start realizing that RAW is not a necessity.

[MiL0]

where has it been confirmed that 46Mbit 1080p has been tested and is working?

all I can see from that video is that someone has managed to decrypt the firmware and view the source files... I think there's still a LOT more work to go before this hack results in anything usable.

[Andrew Reid]

There is a lot more work to do, yes. It was someone else who claimed the 46Mbit, so far not much more info is out about that. I prefer to concentrate on needs of the project, and achieving goals as laid out by Someone 1.00 so please keep the topics on that...

[sanveer]

The FS100 shoots at 24Mbps. I don't think its the bitrate, as much as the kind of compression, and the codec for lower end cameras, which,somehow removes details, for same size sensors. Though, for the GH2, people have been able to get over the problem, and will be able to do so, for almost all cameras (presumably), by increasing the bitrate and tweaking a little bit of the settings, i am sure, there must be some way to keep the bitrate low, and still maintain enormous details and DR.

[Andrew Reid]

The FS100 has an implementation of AVCHD that is as good as it gets, for better you definitely need to double the bitrate no question.

By the way it isn't so much 24p at 24Mbit which is the problem on the FS100 but 60p at 28Mbit, because bitrate is allocated variably per frame. The more frames the more the available data is thinly spread across them.

[Axel]

[quote name='EOSHD' timestamp='1344776311' post='15393']By the way it isn't so much 24p at 24Mbit which is the problem on the FS100 but 60p at 28Mbit, because bitrate is allocated variably per frame. The more frames the more the available data is thinly spread across them.
[/quote]

That's not correct. With Mpeg4, practically all redundant data over time are reduced. The lower the frame rates, the more radically the content alters from frame to frame, so you'd have less redundant information and need higher bit rates. For 120 fps, you'd need about 30 mbps. Ever did stop motion animation? You only draw the changes, and what changes between two frames of 48 fps is a fraction of what changes between two frames in 24 fps in the first place. In hand-drawn animation, you'd simply draw a few motion blurred objects connecting the key frames, no matter how many phases the frame rate actually demanded.

What about camera movement, when [i]every[/i] pixel has changed position in the next frame? That used to be my argument, but the engineers' answer is: If the position changing is predictable, it is a classic motion path, which doesn't even need particularly high bit rates. If it is chaotic ("stressing the codec"), the encoder tends to simplify the details, which by their nature then cannot be recognized as details by the human eyes.

Noise, though not detail, is not only redundant. We perceive noise subconsciously as a natural phenomenon. Watch the shadows below your desk in the evening. Are they dead?

That's why too low bit rates are a problem - at any frame rate: They iron out the random noise. With 60p you [i]always[/i] see this effect, with or without interframe. That's because you have more than twice the temporal resolution. 24p is a very effective way of temporal compression. As little as possible, as much as needed.

[Andrew Reid]

Why does each frame in 1080p/60p look more compressed than the same frame in 24p then?

It's not just about motion estimation but the amount of detail per frame, and more frames = more data.

[Axel]

[quote name='EOSHD' timestamp='1344782683' post='15400']
Why does each frame in 1080p/60p look more compressed than the same frame in 24p then?

It's not just about motion estimation but the amount of detail per frame, and more frames = more data.
[/quote]

I see. I forgot you shoot higher frame rates only to interpret them as 24p and generate slomos. Then you are right.

The consequence is, we need intraframe or at least very small GOPs.

Take an extreme slomo, like 1000 x. All you needed were two frames for Twixtor or Pixelmotion, but those would have to be excellent. No noise, no motion blur. But surely you would have to add noise afterwards, because our air also is filled with, eh, [i]something[/i]. If you had shot the scene with 24.000 fps - intraframe - , you would also not only have the main object changing from frame to frame.

[Andrew Reid]

Here's my understanding of it in a nutshell and it is really quite simple. Bitrate is measured per second, so the more frames per second the more compressed and muddy each will be unless you use I-frames and higher bitrates.

B frames and 'magic' can only get you so far.

Sure you cam shift the bitrate allocation about, right down to even macro blocks within a single frame but eventually you will run out.

Plus H.264 has a habit of shifting too much allocation away from areas of the image like the sky, causing banding. GH2 hack really helped with that.

[christianhubbard]

This thread says tested and working, but I have yet to see any footage...

[galenb]

Very enlightening conversation about bitrate. thanks guys. I'm not sure which of you is "right" though. Both sound conceptually solid. ;-)

[richg101]

[quote name='christianhubbard' timestamp='1344796867' post='15414']
This thread says tested and working, but I have yet to see any footage...
[/quote]

moaner. whether it works or not probably won't affect you anyway. The higher bitrate hack is simple to the guys who did the GH2 hack.. it's the amazing job of cracking the decryption that is big news here - this is proved since he has offered the modified firmware on this forum. The decryption is what has haulted any development so far. If another person on another forum says they have a friend who has hacked it, I believe them. Why lie about it? If one guy proved he can crack the decryption, why can no one else do it independently for themselves, going one step further and writing a modified firmware?

[kmlo1984]

please tell me what is BMCC and FFD?
thanks
K.M.Lo

[Bruno]

[quote name='kmlo1984' timestamp='1344849071' post='15483']
please tell me what is BMCC and FFD?
[/quote]

BMCC = Blackmagic Design Cinema Camera
FFD = Flange Focus Distance

[Bruno]

[quote name='richg101' timestamp='1344818886' post='15458']
If one guy proved he can crack the decryption, why can no one else do it independently for themselves, going one step further and writing a modified firmware?
[/quote]

Because it doesn't mean the hardware can take it.
You could hack the firmware and tell it to shoot 1000 frames per second at 4k, this doesn't mean the camera hardware would be capable of doing it. This camera has enough heating issues as it is, so yeah "Tested and working" is highly misleading and sensationalist at the moment.

[Andrew Reid]

Quick update.

Looks like we have found a member of the Magic Lantern team to oversee the donations. Big thanks to Malcolm for that.

Once that is set up at PayPal, I'll put a donate button on the forums and once the funds reach the required amount our hacker will be getting his test bodies to sacrifice on the alter of NEX hacking :)

[richg101]

[quote name='EOSHD' timestamp='1344866053' post='15513']
Quick update.

Looks like we have found a member of the Magic Lantern team to oversee the donations. Big thanks to Malcolm for that.

Once that is set up at PayPal, I'll put a donate button on the forums and once the funds reach the required amount our hacker will be getting his test bodies to sacrifice on the alter of NEX hacking :)
[/quote]

cheers for the plugging of this Andrew. It's great that a system has been assembled to allow us to contribute to the cause. I'm sure your front page article will bring more than enough backers for Someone1.00 and anyone else that comes in on the project

[sdpizzo]

Very Happy to support this project, just donated $35 (I own both a 5N & a Nex7 so this is awesome), btw, Im ALSO a supporter of Richard Gale so my Kick ass Nex5N with housing is gonna kick some serious butt!

[christianhubbard]

..still havent seen any footage of this tested and working.

How can anyone be expected to donate if it looks like you could be lying right out of the gate?

Not saying you are, I just don't understand why there isn't any footage available to view.