Worried about depending on RED codec

stokestack

stokestack

Active member
Hi all.

If the RED codec is wavelet-based, I'm not sure when you'd want to use it. Wavelet codecs are lossy, which to me defeats much of the purpose of this camera. Fortunately, it looks like there will be plenty of options for output, but if the affordable storage options are only practical when using a lossy codec, that's a potential problem.

Granted, even the high-end CineAlta is mildly compressed. But there's been talk here of data rates like 150 megabits, which for the higher-res modes of this camera seems silly.

Thinking it through practically, the most important time to have pristine 4:4:4 is for greenscreen work, and that's usually done in the studio where portability isn't so important and you can capture to some big AC-powered device. If you're outdoors on location or in a vehicle or what have you, maybe a mildly compressed originating format would be OK. But there are still instances when you block something off on location with a greenscreen, and you'd rather not have a massive external device.

Eventually it would be nice to hear more about goals of the RED codec development from the Red guys. What kind of situations do they expect it to be used in, and for what will it be unsuitable?
 
Last edited:
If compression is visually lossless and holds up to compositing without producing artifacts, does it matter how it works or what the bit rate is?

We can't change the laws of physics for the amount of data uncompressed video takes - the only way to reduce that cost is via a high quality (lossy) codec. We're picking the best we can!

I think the codec will be totally usable with superb quality for lots of users. The only times you'd not use it would be - going out to a deck, or those who want "perfect" quality for sfx shots, or for severly overcranked video.

Also, don't worry about bit rates - we have enough. There's features of the sensor that can give us great compression efficiencies.

Graeme
 
David Newman's Cineform 10-bits codec is well-known for its lossless quality and it is the Silicon Imaging/Cineform camcorder basis.

So, regarding the RED codec I ask if will it be as good as 10-bits?

Following Cineform or it's not possible to expect the same (at least!) from a wavelet-based codec?
 
good

good

Thanks, Graeme. I was hoping you'd weigh in.

"If compression is visually lossless and holds up to compositing without producing artifacts, does it matter how it works or what the bit rate is?"

Nope. But these days there are lots of claims made about codec quality that don't hold water for any discerning viewer. Look at what's being pushed as "HD" today: just about anything.

Specifically I was wondering about edge artifacts for keying. But by saying that the codec holds up to compositing, I assume that you've considered this. Good news!

Gavin
 
Last edited:
Graeme as we all know is a master of codecs.. im sure he will fight tooth and nail to get the best solution possible... his name is on the line :)

Remember, its not political for RED, so they can start from scratch or bring in whomever they want to, without being locked to a specific codec such as DVCpro HD or HDcam etc.
 
Jannard said:
That's why Graeme was added to the team. Well, at least one of the reasons...

Jim
Click to expand...

ha ha yes... plus everyone knows every allstar team needs at least one Canadian :)
 
I don't see the advantage of having the HDV codec as a recording option for RED ONE. If a highly-compressed GOP codec is really needed then I would rather see RED develop their own codec instead of using HDV. But I really can't imagine many people would ever use it - especially with hard drive capacities increasing and costs falling - there would be no point of high compression.

DVC Pro HD is pointless as RED CODEC offers a direct replacement.
 
Martin Hill said:
I don't see the advantage of having the HDV codec as a recording option for RED ONE.
Click to expand...

Amen...Hdv should never had existed.
Heres what it does, I get a red camera, and say to my clients that I can shoot Hd. They say, "big deal, I have my ...(fill in blank)... camera here".

Not everyone, especially paying clients (if your not in the film industry) keeps on top of this stuff.

Anyway, the topic of codecs does bring me to workflow.

If I capture with the red codec, and then I digitize (well I dont digitize now do I...drag & drop?) anyway, its in my FCP station using God knows what graphic card (Kona I suppose)...what do I send out to? Sure if the end is a dvd, I can give that to a client. But if they order Hd 4k, do I give them blue ray? sorry, a bit confused. Send it to tape? (dvcproHd or what?)

And if they need a back up, I suppose I have to give them a dlt.
But then they say, I cant edit this at so & so's place because they dont have a kona card & support for the 4k Raw or the 2k codec, or cant read the dlt, etc., etc. (I say, stick with me...hehe) But realistically, Im trying to figure out this whole end product deal. What does it look like once shot, even after I bring it in?
Not only for how I store it, but how others can have it and work with it, WITHOUT me having to buy a dvcpro 50 hd deck...see this is where it gets interesting. I cant, or rather wont put money out for a deck...the point is tapless after all.

Well, now that is as clear as mud, :) but Im sure you guys can feel you way through what I said.

Look forward to some of the comments.

peace

dalen
 
Last edited:
There's no advantage having either HDV (way too compressed, doesn't support 60p or 2k) or DVCProHD (old, DCT based, 8bit, proprietary) on board when we can go with a modern codec designed for the way that digital cinema cameras work, not something that's ok for ENG.

That said, I'm not philosophically adverse to the notion of a high compression long GOP codec on board for when record time matters more than absolute quality - it's just that that's not what I reckon people on this board, or myself want in the camera. I'd just pick something more modern than MPEG2 though.

And people know that I'm not accepting of poor quality in video codecs. I'll make sure it's not just visually lossless, but stands up to a good degree of compositing also. Both HDV and DVCproHD can't stand up to strong colour correction, at least, they can't when I try it.

Graeme
 
dalen7 said:
But if they order Hd 4k, do I give them blue ray? sorry, a bit confused. Send it to tape? (dvcproHd or what?)

dalen
Click to expand...

As cheap as they are, couldn't you just give them a hard-drive? Just figure it into your cost. I suppose you could even do a value-add and give them an external drive.

Isn't that how Studios were looking into disti. for the newer digital features?
 
Graeme, will a 10 bit codec be sufficient for heavy CC? Or is the bitdepth less important than the codec itself?
Is there a plan to implement more bitdepth?
 
JavaDog said:
As cheap as they are, couldn't you just give them a hard-drive? Just figure it into your cost. I suppose you could even do a value-add and give them an external drive.

Isn't that how Studios were looking into disti. for the newer digital features?
Click to expand...

I guess you could...how would they edit it? Guess they would require a robust set up or depend on you to do it?

peace

dalen
 
DVCPRO-HD can't really take any color correction

DVCPRO-HD can't really take any color correction

I was shocked the first time I started looking at Varicam footage really closely. There is tons of banding and noise in the dark parts of the image. Even the slightest HSL tweaks in color finesse creates a lot of visual junk.

HDV on the other hand is just pure dooky.

Two main thing to think here:

1) Will my favored NLE support RED at some point in the near future?
2) Will it look better than HDV or DVCPRO HD?

The answers:
1) You have Ted S. from AJA in there who has just a little bit of experience on the desktop video front and probably a few contacts as well. Based on the NAB reaction you KNOW every NLE will want be supporting this camera.

2) Frankly, it shouldn't be hard to make a better compression scheme than HDV or DVCPRO-HD. Also, have a little look at www.nattress.com and get a sense of the quality of thinking Graeme is known for.

-RED 142
 
The good news about the REDCODE codec is that it will be upgradeable. As it gets better, you will be able to benefit from the latest versions without buying a new camera.

Also, the REDCODE can be a proxy to RAW when recording via the RAW port or to other formats when recording via HD/SDI since we can record internally /RAW & HD/SDI simultaneously.

But ultimately, we beleive that when people see the quality of REDCODE and the seemless workflow it will offer with NLEs, the majority will choose to stay in REDCODE (10bit 4:4:4).

Wavelet based codecs (Cineform) are used today as DI solutions. Meaning, film scans are generated based on that codec. That's very impressive for the efficient bitrate they offer.

Also, RED is an open architecture. It has been the philosophy since conception. We are going to welcome third party companies to develop RED accessories. If a company finds that there is a market for HDV or other Long GOP recording format for RED, they might develop their own RED Long GOP recording solution or not so long GOP ;). We will welcome it with open arms.

As far as bitrate, the plans so far is to make REDCODE a variable bitrate codec. It is a very efficient way to go. You don't waste bandwith for scenes that don't need it.

Bottom line... under Graeme's leadership we believe that REDCODE is going to be one of the best, most efficient codec available.

Frederic
 
10bit is enough for good colour correction yes. More bits are better, but video is actually worse than 8bit (even though it's 8bit) because it's YCbCr not RGB. It only uses levels 16-235 not the full 0-255 for instance.

Both bit depth and codec are important for colour correction. If you did DVCproHD in 10bit, it would still be bad due to the compression artifacts, and that it tends to compress darker regions more than lighter ones, and hence puts artifacts into the shadows. It's really designed for ENG where you don't colour correct much - it's pretty to the eye on first decode, but after that degrades seriously. When I'm testing codecs, I look at them visually, and then put a colour correction torture test on them to see what breaks up.

Graeme
 
As long as the codec can be easily edited in Avid, FCP and be able to be taken to a high end finishing suite for filmout (like the DS Nitris), I'll be satisfied.

I'm not to concerned with green screen work, but more with achieving 35mm DOF, latitude, FOV etc... I think 2K 10 bit should retain those characteristics in even an average codec correct ? Give me that aesthetic, so I can create a lush visual poetry similar to film, at this price point, and I'm chuffed.

BTW - you'll probably have to transcode to DNxHD to edit any of the RED footage on Avid anyway. Maybe their codec (except it's acquisition) is similar to the Avid one, to allow huge amounts of data to be collected/edited without losing the essential info...:beer:



Cheers.
 
Last edited:
D_and_G said:
As long as the codec can be easily edited in Avid, FCP and be able to be taken to a high end finishing suite for filmout (like the DS Nitris), I'll be satisfied.
Click to expand...

Absolutely! This has been one of our top priority since inception. Everyone on the RED team starting with Jim agrees that by time of shipping the post workflow has to be seemless.

We've all suffered (and some of us benefited;)) too much from cameras with formats that aren't supported by the top NLEs for months sometime years after shipping.

Frederic
 
You must log in or register to reply here.
Back
Top