Redray for the scarlet, as low end solution and for HDTV?

[Wayne Morellini]

I have a suggestion, why not include red ray on the scarlet (though I imagine you guys are already planning this). As a low end pro video solution, instead of h264 inter cameras.

Most video work is done for TV, maybe even eventually blu-ray as well, and it is a huge cost and waste of disk space to record at the data rate of red code for the quality of TV. From the sounds of it, Red Ray greatly surpasses the quality of HDTV and can match blu-ray at much lower data rates. So, for low end people to be able to use red day for recording as well distribution makes sense. It also makes sense to promote it as a HDTV standard.

If red ray truly as good as people are making out, few of us could could make a high compression codec that good, then it also makes sense for digital cinema.

I know this is not the normal, build on what's out there, conformist, thing I might say, but I see that Red Ray offers a great advantage here.

---

Re-edit: I forgot to include, Redray's advantage as a Super and Ultra HDTV standard, and the possibility of including that in a low cost fingernail sized solution in coming years in partnership with an existing low powered codec processing firm that I know.

 

[Anthony Rinaldi]

Wasn't there talk of in-camera debayering? Something like a Red Rocket via the Camera?

 

[Wayne Morellini]

Still, only one reply. I hear that Apple is working on a alternative codec to H264 (redray?). Microsoft still has not signed to BR in it's xbox (though the DVD capacity is limiting).

Could I further suggest, red ray for consumer HD and recording devices would be a major advantage (even for SHD TV transmission). There is one company I know of out there with the expertise to put a codec like redray into these devices, even into pocket platforms, ambarella.


Thanks


Wayne.

 

[Pietro Impagliazzo]

Do we even know how much processing power it takes to encode REDRAY? And to decode?

Highly compressed is not the optimal solution for capture, not in our current time.

GH1 is hacked and get better results... When the bitrate is elevated considerably, when MJPEG is used instead of AVCHD.

I'm not sure the current Scarlet camera design can take in-camera REDRAY encoding... But I do know that a low bitrate REDCODE would be an awesome solution, specially for 3K shooting.

However, REDRAY as a storage solution would be an excellent addition.

 

[Wayne Morellini]

Do we even know how much processing power
it takes to encode REDRAY? And to decode?

Yes I see what you mean Pietro, it might require moire processing power than the existing raw format.

But, how do you know how much processing power it takes to encode red
ray? How much do you know of what I know?

It is believed to be demonstrating figures, that are similar figures to my
own previously planned codec out of date compared to my present concepts. Normally I would regard it unlikely, but my estimate of the person behind this is that he is the 'type' of person that
could do it.

If the technique is done well it will require little processing power
compared to the intra codec. Low compression can require a lot of
processing merely because of the larger amount of data, while you can
design a high compression codec to progressively throw out data at the
start. Some other techniques though, give it the quality. So, before
posting I considered the likely hood that Red ray might actually require
less processing power, and therefore fit in the camera.

Highly compressed is not the optimal solution for capture, not in
our current time.

Doesn't that depend on the application? Here we are merely talking about
replacing the large number of prosumer h264 inter codec cameras and the
professional h264 intra codec cameras, both of which, regrettably, produce
better compression quality than the Mpeg2 based HDTV systems. So, by the
reports (as mistaken as they likely to be) Redray is likely to eb more
than enough for the application, even if they move to h264 inter HDTV.
What is more, they are moving to Super HDTV and Ultra HDTV in future, and
that is where Red Ray can prove a real asset compared to the quality you
will get from the wider channels in h264. But on the capture side, it
merely makes a much more achievable system for those shooting footage, and
maybe able to shoot up to the standard of 100mb/s intra given enough
bandwidth.

GH1 is hacked and get better results... When the bitrate is
elevated considerably, when MJPEG is used instead of AVCHD.

Has the camera got the new version of jpeg that competes with H264 intra
on quality now? I thought the bitrate was also much lover than needed (but if
it can do 200-400Mb/s, we are talking (actually there is a cheap fuji camera that reportedly does more than 200mb/s I read in the specs). I am too waiting to see what the GH2 brings, apart from possible 3D lens and fullHD 1080p60.

However, REDRAY as a storage solution would be an excellent
addition.

Yes it would, in a way.

I posted here, as I often do, to prompt thought on new features, and there are a list of them included, and for rational debate on the subject. If people can see the benefit, it is up to them to put their preference in for the feature to be noticed by the higher up powers and considered for only possible inclusion. I consider what I post, how practical, needed and good it is for some group of users, and how right it is, before I post a thread. All this ignoring and illegitimate stomping on people that goes on around the forum, just hurts the camera. If we relied on that, we would HAVE to be content with something like Box Brownies by this time in the industry. If we were more content with the other side, global warming would be solved and we may have reached the stars by now.

 

[Wayne Morellini]

I forgot to mention the possibility of including Redray in a low cost fingernail sized solution in coming years, in partnership with an existing low powered codec processing firm. The technology is already in use, and in coming years the shrinkage of chip process technology, as well as IBM and Intel plans for new types of chips, will greatly increase the amount of processing power. I do not know about the IBM/Intel time frame, but shrinkage of process continues. What is interesting, is if a cheap enough solution can be had by the time SHDTV or UHDTV broadcasting goes mass market. If redray is implemented efficiently, we could even see a cheap solution by then. Before then there is potential for a disk/download to SHDTV market.

 

[Stephen Gentle]

If red ray truly as good as people are making out, few of us could could make a high compression codec that good, then it also makes sense for digital cinema

It doesn't make any sense at all to shoot on. Apart from the extra processing power it would require (which would mean a bigger camera, shorter battery life, and it would cost more), one of the thing that codecs meant for distribution do to pull down the data rate is to throw away as much information as they can. That's fine normally, when you are just watching it, but it is exactly the information you need if you want to do any sort of grading on the footage without it falling apart.

I forgot to mention the possibility of including Redray in a low cost fingernail sized solution in coming years, in partnership with an existing low powered codec processing firm.

RED (probably) already have silicon to decode REDRAY video, but I don't think they would have bothered making ASICs to encode it since it's not suitable as a format for shooting and the RED ROCKET and a regular CPU in a computer probably does a good enough job at it.

 

[Wayne Morellini]

It doesn't make any sense at all to shoot on. Apart from the extra processing power it would require (which would mean a bigger camera, shorter battery life, and it would cost more), one of the thing that codecs meant for distribution do to pull down the data rate is to throw away as much information as they can. That's fine normally, when you are just watching it, but it is exactly the information you need if you want to do any sort of grading on the footage without it falling apart.

Hi Stephen, you might like to read what I have posted, I have already covered them. We are talking about the cheap and nasty highly compressed mode alongside red code, to replace prosumer h264 cameras. This would be a few grades above HDTV, (if not at BD datarates) so ample room to work with for those people. We are not talking about this mode for the film market, but the much bigger ground based TV market.

RED (probably) already have silicon to decode REDRAY video, but I don't think they would have bothered making ASICs to encode it since it's not suitable as a format for shooting and the RED ROCKET and a regular CPU in a computer probably does a good enough job at it.

They probably do, or at least a FPGA design, both of which could be transfered over to the scarlet, especially if it features a FPGA processing engine in there (that can be reprogrammed with a data file at any time back and forwards). There are also other types of processing system that could allow a redray mode to also be added after market as a firmware upgrade. It is a perfect bone crunching mode for the fixed camera (even h264 is) against other cameras in that segment (one of their biggest remaining advantages is that they have small industry standard codec files to hand over on jobs).

As for the PC, they definitely want something better than that to decode red ray. I think it unlikely they can get bit on a PC lowcost and low powered enough to go into low end mass produced products. Much of what you see in consumer video does not rely on the PC, much is custom chips or lower cost to power processing forms. An example of a cheap implementation of redray might be implemented on a GPU with general purpose processing abilities, programmed in OpenCL (Open Computer language for portable parallel processing code across multiple processing engines). This would enable much of the specification to also be portably shared with the PC, Mac, game consoles and other processing platforms, reducing the cost and increasing the flexibility.

 

[Kaku Ito]

Hello Wayne, long time no talk.

Have you had chance to shoot with RAW? Now with STORM coming up, you won't even think to shoot in other formats. I think RED is coming out with a Prores module for quick editing, but even though. I'm confident that I can handle any TV drama job with my setup any time at this point.

 

[Wayne Morellini]

Hello Kaku good to talk to you.

Different quality for different jobs. I am neither possessed or obsessed about raw or the scarlet, I might even land up buying a Panasonic GH2, and the information I read after posting is most promising.

The quality would not be such a issue for most in field production work, due to the limitations of TV and BD. It would be much more convenient with the reduced size and cost of 20-50mb/s, for the majority of smaller players where sales are. I am going to suggest 10 bit as well.

I have a Sanyo CG200 HD pocket camera for personal use, and even 17mb/s datarate (24mb/s for older camera) is a pain with the SD cards (HDD option would solve that). One day a codec of redray's performance should be able to fit in such a camera with the same type of re-programmable multi-processing core array it uses. Alternatively before then, I would hazard to guess, it might even be possible to program such a codec with the GPU of the Iphone 4G or future 5G.

 

[David Rasberry]

For the TV crowd quick turn stuff, H.264 makes the most sense currently. If not from Red you can get a real time broadcast grade wireless H.264 module now for about $2500.

Red ray as I understand it is strictly a release format. Not suitable for editing.

 

[Kaku Ito]

RED IS coming out with H.264 wireless module for Epic, so I'd figure you could use it for Scarlet, too. Also Scarlet does the 1080 resizing recording, so it would be useful to broadcast people. Overcrank ability must be really useful for extending the expression even for the TV shows.

R3D being really small, I would never bother to shoot other than RAW. Even with NEX5, when you have super white and super black, you are glad that you had the range, so shooting with RAW and have more range to fix or express more in grading, also never can make mistake in white balance.

AVCHD camera with DSLR lenses are good, especially when they are small like NEX5 (too bad it does not do 24p) so I can take it around any time. I'm pretty sure if Wayne put his hands on real Scarlet, he will just start adopting the RAW ideas. Then if Jim comes out with real pocket camera with RAW recording at the end, I'd be so stoked.

 

[Wayne Morellini]

Am rushign out the door, so breifly.

I am actually talking about a higher datarate gradable Redray in 10bits. Similar or better in quality to h264 Intra, but at h264 inter-lowend H264 intra bitrates (20-50mb/s). I am talking about for low end private jobs, news reporting and interviewing, the bulk of the work. I was also talking about for SHDTV broadcast in 10-20mb/s. For high end jobs like TV drama, or movies, I also would prefer RAW myself. I myself was a part of the group that led to the Red and Raw revolution, before I turned my attention to increasing the quality of smaller cameras over at dvi, so I have a appreciation of Raw.

Have a good day.


Thanks

Wayne.

 

[Kaku Ito]

Sorry for not being aware of you're being aware about RAW, but now with STORM that takes advantage of RAW on Red Rocket OR GPU (doesn't even have to be CUDA) to get realtime performances to edit or process, discussing lossy H264 as a recording format is not really that meaningful, that is my opinion. I do have a NEX5 that shoot AVCHD and it really wraps up nice image than Canon EOS in my opinion, but still, AVCHD is very lossy format. I'd say GPU to be able to play R3D (especially with HD size) will be possible very soon, so instead of spending efforts on making the distribution format to be recording format, I'd spend the recording format to be more understood and accepted more.

 

[Wayne Morellini]

The distribution format has become the same as the recording format already. 3DTV is moving towards h264 used in AVCHD and bluray, and their is a market for TV stations to store their content in h264 instead of old mpeg2.

But again, we are not talking about the same sub-format of RR, for capture we are talking about a higher data rate, increased bit depth and provably 4:2:2, that is a fraction of the data rate of AVCHD intra for the same quality, and less than the inter version for the same quality. For transmission we are talking about something similar to present RR, but with encoding for broadcast transmission, similar to present allowable HDTV streams. So, the only thing to compare it with is the lower compressed professional intra version of h264. However, if RR is doing its trick by elimination of detail and quality, than this may not hold true. As I demonstrate to industry professionals in the past, high compression can be achieved through elimination but is not really meaningful. I could compress a given screen with as many pixels as there are particles in the universe, in a single byte, even a bit. That image is of course, a blank screen, the compression command is to render one, but it is [perfectly useless in quality for anything else (Black screen, white screen). So, eliminating differences in the image to increase compression is not good for real quality, only good for increasing perceived quality in certain ways, and mostly useful in sacrificing parts that won't be so noticed. Years ago I came up with multiple compression techniques of similar compression to what they are claiming, some could do this through elimination, and others would do this with minimal elimination. I hope they are doing the latter.

We should not get too far into the GPU thing, OpenGL is a valid open format for this, that covers non-nvida GPU's. There are certain intensive things that happen in low compression intra codecs that make them processing hogs, though I do not know how it transfers over to GPU processing. The mere elimination process in inter codecs apparently reduces processing loads (remembering at the same time complexity of schemes increase processing loads). So the GPU for inter might be less and available sooner then intra. But the truth is that I calculate that you can do intra more efficiently then presently and more then present inter (but also inter could be done more efficiently too, which wins I don't know yet).

However, flash is running out of shrinkage room, and we are planning to move to newer technologies eventually. But the issue is that our lives, media files etc take up a lot of room over a lifetime, and better compression is still practical.

At the moment, for instance, I am looking into setting up a cheap system to record 5 HDTV network streams simultaneously 24 hours a day, then to watch what I want and eliminate the rest every now and then. A great product that can be widened to however many stations are available in your region. I've mentioned this to leading h264 provider before, but redray would be even better. So, RR has a wider market there that could be exploited, and why not. I am so sick most of the time I can't implement my pre-RR codec at the moment, but if they can do it, and the community get the benefit out of it that is great. In the end it is not about ourselves, and what we can individually get out of it, what it can do for other people is more important. Do you agree, Jim? Even this codec, if as good as I hope, is still magnitudes off the mark of what can be done.

 

[Kevin Olsen]

Wasn't there talk of in-camera debayering? Something like a Red Rocket via the Camera?

Ya, I started a thread about this, using the in-camera processing engine to do the debayer while connected to a computer, but no one really seemed to get behind it.

In my opinion, for Scarlet users especially, this feature would be exceptionally useful and I am still a little surprised that it went unnoticed.

Hopefully we can get some more traction on this idea!

 

[Wayne Morellini]

Thanks Kevin,

Yes, I have offered such thought before, mainly to a h264 manufacturer, as the footage was hard to play or edit on PC's then. It is a good step around problems, especially if USB 3 could be involved, until PC's catch up. Infield it is also a good option to cull certain things without having to wait till you get back to your computer.

This is about the major low end market, but people tend to see things in the way it affects them directly, their experience of what has been done, rather than what can be done or other peoples use, and it requires some thought. Hopefully the people in charge, have been watching, as with many times before, and these things are being taken into account with as much consideration as they deserve.