Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

2K scaling...

Status
Not open for further replies.
Depending on how the sensor technology evolves, I would much rather see pushing towards greater dynamic range, less noise, etc.. before we see a boost in resolution over 4K.

Oh, and I'd like my 4K to some day do 1000fps. But I'm greedy too.
 
I agree that the ability to record longer with 35mm DOF is very important for me, and what I do for lots of my clients. My Red One will rarely be used for theatrical shooting. It's a potential replacement for my HDX-900 with Pro-35 for use in commercials, music videos, and corporate films. However, the onerous rendering of 4K makes it problematic for lots of my clients. Most of them want to shoot tons of footage, and want to walk away at the end of the day with files they can use right away. I record to both the Firestore and to tape as a backup, and it's a fast and reliable system with a minimal amount of fuss on the post end. As it stands now, I can almost shoot film and have it to them faster than rendering Red files.
I suppose I could buy a Panasonic 1400 deck for $25,000 and record the output from the camera, but BOY does that seem like a step backwards.
Ideally the camera would scale to 2K, or 1080, or 720, onboard, in real time. I'd even pay MORE for a Red One with a more robust FPGA to allow that...
But what about this instead? A Red Decoder Box. Some standalone device with hardware-based storage and up-, down-, and cross-conversion. Maybe you insert a CF card, or plug in a Red Drive, and at 4X real time (maybe?) you can scale your glorious 4K footage to whatever you want. Make it crash-proof, bullet proof, controlled by an external PC or Mac.
I'd pay some real dough for that.
But I'm sure everyone at Red is pretty busy right about now... :)
Cheers,
Harry
 
How about 4k at 60fps AND 2k scaled at 30fps.
I would like to chime in with the other documentarists. If you have to shoot day and night for an intense 3 week period 2k scaled would save a lot of boring late night conversion time! shoot, transfer, sleep is alot better than shoot, transfer, scale, wait. Secondly I think that without 2k scaled, it will become neccesary to bring a desktop, for the ekstra processing, which again makes it hard for doc-shooters on the move.
Thirdly CF cards are GREAT for unobtrusive shooting, crowds, live events etc. with the battery in the pocket. But 8 or even 20 minutes is too little.
and then off course I would second the storage and post production benefits.
4. I don´t see how the argument about higher capacity CF cards coming out apply, because as a documentarist you often need all the storage you can get, which means that higher capacity CF cards would just make 2k scaled an even greater option than it already is! (should be).

And frankly I dont see how the possibillity of scaling immediately can be seen as a compromise. It´s frigging awesome not having to do it later. And since 90 percent of everything shot on RED will be ending up in 1080 or less, this is an issue with serious timeconsuming consequenses for thousands of people.
I know that on many boards it´s preferable to take the standpoint of a "big bucks" DOP, but RED is also for the little guy working his ass off trying to make ends meet... So let King Solomon bring in the 2k scaled at 30fps option :sorcerer:
 
I think the compromise to achieve 2K scaling from a 4K RAW file should be obvious -- you're not talking about simply downrezzing 4K RGB images to 2K RGB images. In RAW, the color information and detail is still tied up in that Bayer pattern, so simply deleting every other photosite's info isn't going to work, and converting to RGB, resizing, and back to RAW doesn't make any sense either. Remember that conversions to 1080P and 720P are either 4:2:2 YCrCb or 4:4:4 RGB, you don't have to end up as a RAW file again.

Like I said, no one would need 2K scaled if they could record 4K RAW at the sorts of speeds and lengths they want, and convert to RGB at the speed they want. Afterall, you have to convert 2K RAW as well, so what's wrong with 4K RAW conversions IF they could be done faster? That's what I mean by compromise -- 2K scaling is just a temporary solution before 4K RAW becomes more manageable.

Besides, for a doc, would you perhaps be better off with 2K cropped rather than deal with 35mm depth of field on a documentary?

I understand the attraction for 2K scaling, 35mm depth of field in 2K RAW, but I wonder if it will be a perfect conversion, RAW to RAW, or if the artifacts from doing the conversion in camera will negate the benefits. I don't think it will be a lossless conversion (obviously, resolution-wise, but I worry about other artifacts).

Besides, if converting 4K RAW properly in post is time-consuming, why should it be easier for a camera's processor to do it faster than a larger computer?
 
Nice one Eddie, I could not agree more.

I don't really mind if the higher framerates cannot be recorded scaled. That is probably just not possible for now and at least good alternatives are offered on this for times you really need it - like the 4K 60fps. However the need to simply shoot 2k scaled at a normal rate of 24/25/30 fps is quite big for us. That would simply be what we would use all the time. And I think that should be possible right? You guys fixed the cold issue in a matter of days as well :)
 
Yes, you can't do scaling by throwing away samples. You've got to filter the image first to remove any detail that's "not allowed" in the smaller image, or else aliasing occurs. A good downsampling filter that gets nice resolution, and a sharp image with very little aliasing and no ringing is hard to achieve. You've really got to work well to make it happen.

Graeme
 
The need for Speed

The need for Speed

This discussion has managed to chase its tail far too long. Perhaps I can help clarify:

Scaled 2K created by the camera would be a great option but AFAIK that is not an attainable spec due to the processing requirements. IF it takes a desktop computer several times real time to do the scaling how in the heck is the camera supposed to pull it off in real time?

I have posted before that based on current specs the killer strategy for docos is to use windowed 2K with either an S16 zoom & microforce or HD glass and a B4 adapter if you don't mind the light loss.

Good news: from what we have heard the roll time limitations of shooting 4K are a temporary issue. Larger CF cards, RedDrives, SSID based options via eSATA, etc are all real possibilities and at least one of those should be live before NAB.

Better news: you can use proxies for editorial and only render what you actually use. Rendering times will get shorter as the code gets optimized and faster hardware ships. Would you consider an overnight render of 8 hours to spit out 1 hour of finished product reasonable? I would and I expect to have that by NAB 2008 assuming I have a fast machine.

Bottom line, you have a 2K proxy generated in camera from the 4K frame that plays back with no rendering right now. What exactly is the problem?

Moreover, if you did scale to 2K in camera then you would never have a 4K asset for future use like you do now. IMHO 28MB/sec of storage and the horsepower to scale it into almost anything in real time will be as easy and cheap as crunching DV is now in just a couple of years. How long before AJA or someone like them has hardware based acceleration on the market for RedCode RAW?
 
Blair S. Paulsen said:
IF it takes a desktop computer several times real time to do the scaling how in the heck is the camera supposed to pull it off in real time?
Click to expand...

Dedicated hardware. The same way the RED camera is able to encode at least one 2k and two 1k channels to JP2 in real time. My Quad Core definitely couldn't even keep up with that.

The problem is a "Debayer->Scale->Soften->Bayer->Encode" chip doesn't exist and would have to be custom designed/fabbed. If you're operating in F35 price ranges that's acceptable. Red has to use what already exists wherever humanely possible.

General purpose hardware is wonderful because you can release lots of updates and improvements after you ship. It's not so good at doing things super fast.
 
It's come up before, but perhaps a dedicated "processing" box is in order. It could essentially be a "little black box" similar in size to the AJA IOHD or of similar nature, using many of the same internals that are used in the RED One. It could be designed to playback REDCODE at 4K in real-time, transcode it on the fly to a few common formats, scale to desired resolutions, etc..

OTOH, I'm thinking that common PCs will be capable of this sort of thing within the next 2 years. A lot will come with continued code optimization and leveraging GPU and other accelerators. The physics / environmental accelerators that are now showing up in gaming systems are another potential, un-tapped resource for processing video.
 
Yes, imThatOneGuy. DEDICATED HARDWARE. Asking a general purpose computer to do this stuff, no matter how amazing personal computers have gotten, is not a professional solution. I'd love to have the flexibility and power to transcode my 4K Red footage into any common resolution and format significantly faster than real time.
I'd pay another $17,500 for that box.
In fact, I'd pay DOUBLE for the Red Two that would be a more robust solution. I think in some ways Jim & Co. have boxed themeless into a corner trying to meet the seemingly arbitrarily chosen price point of the camera.
Don't get me wrong; I'm a believer, and I put my money where my mouth is. I just think that there are many pros like myself that care less about the price of the hardware and more about the power of the hardware.
Still, the camera is an amazing achievement!
Cheers,
Harry
 
shoot a lot of small budget and doc. stuff slinn's and others arguments for keeping 2K scaled are persuasive to me.

Additionally switching between 2K windowed and 2K scaled with the same lens up would act like a doubler. and the 2K scaled image would probably better match the high speed which can only be done in 2K windowed.

Given all that Blair seems to be saying this is actually impossible for RED to do?

It certainly has a lot of value to me as it prevents the need for a second set of S16mm lenses or an adapter to HD lenses at a lot of extra cost.
 
The most economical solution will be in the mean time to buy a $10,000 render farm.

For $10,000 you could very easily build a 28 core farm to chug through footage. This'll require a command line REDCine though to make efficient and of course not be field portable.

Personally I love the idea behind the new Codex Jpeg2000 box. You hook it up. It records. At the end of the day you can output HD-SDI, Data in one of several formats or the full Jp2k original through firewire. Jpeg2k isn't my preferred intermediate format. Infact I don't like it as an intermediate format since it's so slow but it does have broad support and hardware encoders/decoders.
 
"Dedicated Hardware".. That sounds like Avid.. :usd:

"We shall never cease from exploration, and the end of all our exploring shall be to arrive where we started, and know the place for the first time."
 
Why is it that all kinds of people without RED-affiliation all of a sudden are experts on the what is possible and not. Red is famous for accomplishing great things... and looking at the camera specs I would say that 2k scaled at 30fps does not seem like a mount everest of technological achievements, compared to the rest of goodies packed in there.
Secondly scaling raw in-camera is not a compromise, it is the best of both worlds for anyone "settling" for a quick-and-dirty 1080p broadcast or dvd-out. Massive storage improvement, and a hell of a time-saver.
thirdly, documentaries is not bound to 16mm dof by some law of neccesity, only by lack of funds. Large amounts of 35mm docs is hopefully gonna be one of the great achievements in film history spawned by RED ONE. Here is where people in the future can really look back and say... oooops what a leap. I dont have any problems focussing at f 5,6 with 35mm, especially when it is scaled down, so I have to disagree with the tendency to put down docs as aestetically inferiour or less important, than the big budget "My greek wedding VI."
I mean telling stories is fine... but telling the truth is sometimes even better.

All the best

eddie
 
"Sounds Like Avid"
Yes, and Media 100. Expensive. BUT... they worked and were professional tools; robust, fast, and deep. Then Final Cut came out and every 21 year old film school grad said how it was so great and so cheap. I bought a copy... ingested some footage... made a few cuts... and then went to play back and see how my cuts were, and guess what? UNRENDERED. Yep, every time I made a simple cut, I had to render. BIG step backwards from my (at the time) 7-year old Media 100 system. Now we all know it's gotten a bit better than that, but my point is that cheap and accessible at the expense of functionality that was already available is not always great. And my time is valuable to me.
YES. DEDICATED HARDWARE. YES IT SOUNDS LIKE AVID. But I want a professional tool that I can use today, not some cobbled together render farm in my office. I'm on to the next job after wrap, and I want to hand the client footage they can use.
Sorry if it sounds like a flame; it's not. I just want Red to be all that it was cracked up to be 14 months ago. I WILL pay more. But let's not get the ball to the goal line and say, "OK guys. Trust me. This is what you want."
Cheers,
Harry
 
Big fat maybe

Big fat maybe

My thesis about what is or is not possible is based on Jim's posts, Graeme's posts and common sense. Obviously the Red Team may prove me wrong, so be it. In the meantime I am going to continue to participate in this discussion, you may ignore me if you like.
 
David Mullen ASC said:
I think the compromise to achieve 2K scaling from a 4K RAW file should be obvious -- you're not talking about simply downrezzing 4K RGB images to 2K RGB images. In RAW, the color information and detail is still tied up in that Bayer pattern, so simply deleting every other photosite's info isn't going to work, and converting to RGB, resizing, and back to RAW doesn't make any sense either. Remember that conversions to 1080P and 720P are either 4:2:2 YCrCb or 4:4:4 RGB, you don't have to end up as a RAW file again.

Like I said, no one would need 2K scaled if they could record 4K RAW at the sorts of speeds and lengths they want, and convert to RGB at the speed they want. Afterall, you have to convert 2K RAW as well, so what's wrong with 4K RAW conversions IF they could be done faster? That's what I mean by compromise -- 2K scaling is just a temporary solution before 4K RAW becomes more manageable.

Besides, for a doc, would you perhaps be better off with 2K cropped rather than deal with 35mm depth of field on a documentary?

I understand the attraction for 2K scaling, 35mm depth of field in 2K RAW, but I wonder if it will be a perfect conversion, RAW to RAW, or if the artifacts from doing the conversion in camera will negate the benefits. I don't think it will be a lossless conversion (obviously, resolution-wise, but I worry about other artifacts).

Besides, if converting 4K RAW properly in post is time-consuming, why should it be easier for a camera's processor to do it faster than a larger computer?
Click to expand...

Well yes, David, 2K (scaled or not) is a compromise. But I'd rather make that compromise and be able to use a Red One than having to use another camera just because of storage issues.

Let's be clear, of course I will shoot 4K every time I can.

But what if you leave for 1 month for a doco to a remote location with a (very) small crew ? You will come back with at least 50 hours of footage. That's about 5 320GB Red-drives in 2K, but 20 if you shoot 4K. And this doesn't include backup, which is mandatory if you shoot on harddrives.
Or 50 32GB CF Cards for 2K, but 400 16GB CF Cards for 4K...

So my point is, I could deal with the storage needed for shooting 2K, 4K seems harder to manage in that particular situation.

I'm asking for this because I may find myself in this kind of shooting scenario quite soon...

David Mullen ASC said:
Like I said, no one would need 2K scaled if they could record 4K RAW at the sorts of speeds and lengths they want, and convert to RGB at the speed they want. Afterall, you have to convert 2K RAW as well, so what's wrong with 4K RAW conversions IF they could be done faster? That's what I mean by compromise -- 2K scaling is just a temporary solution before 4K RAW becomes more manageable.
Click to expand...
That's exactly my point !
IF there were much bigger CF cards available for affordable prices...
IF there were bigger / faster / cheaper SSD drives available...
IF computers were faster...
Then 2K scaled would be useless.

BUT for now, and probably at least for the next couple of years, this won't be the case, so 2K scaled would really help in the meantime...

David Mullen ASC said:
Besides, for a doc, would you perhaps be better off with 2K cropped rather than deal with 35mm depth of field on a documentary?
Click to expand...

Of course I could shoot 2K windowed, but why could'nt you have a shallow DOF on docs ? More than once I've shot docs with mini-35 + HVX, and it looked fine. P2 was a pain in the ass, but 35mm DOF looked great and focus was ok. Why should docs always have a crappy video look ?

David Mullen ASC said:
Besides, if converting 4K RAW properly in post is time-consuming, why should it be easier for a camera's processor to do it faster than a larger computer?
Click to expand...

For those saying 2K scaled is not doable in camera, then how comes it is able to debayer and downconvert to 1080P and 720P RGB multiple times in camera ? Most high end computers struggle doing that, but the camera is able to do it just fine. Why wouldn't that apply to scaling RAW ?
But all this is just speculation, no one here knows what can and cannot be done in camera, only Red knows (and even they learn new things every day).

This quote from Graeme in another thread a few weeks ago made me think they still thought it was possible :
Graeme Nattress said:
I understand all what you're talking about. The problem with "simple" scaling techniques, raw or otherwise, is the evil that is aliasing.

I know the need for 2k scaled, for high fps full sensor DOF. I know what I have to do.... That's about all I can say.

Graeme
Click to expand...

In fact, it would be nice having a clear statement from someone at Red (Graeme ?) regarding this, because if it really can't be done, then all this discussion is just pointless...

Nils.
 
Status
Not open for further replies.
Back
Top