Welcome to our community

Be a part of something great, join today!

  • 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 :)

Why do we need a RED Rocket card?

Even Adobe is not making much use of the GPU for decoding, they use it for effects. If you have a really beefy machine, you can work quite well without a Rocket.

Plus, the Rocket is giving you video out, which Adobe can't do when CUDA is active – too much shuffling around of data between the two cards.

IMHO, RED has chosen wisely and is giving you options. I can see no conspiracy here.
 
As soon as there is a graphics card that can do what Rocket can do, I'm in. But right now, all I know is the only way to crank through transcodes at around 50 fps on Epic 5K material is with 2 Rockets which is what we did on a move this summer. They were running 3 cameras a day quite often, and because of the Rockets I could get through this stuff in a reasonable amount of time to create dailies. According to Deanan there is no GPUs out there that could get me this kind of performance, so time is money, and this was a good solution.
 
You can get pretty decent transcode speeds from Davinci Resolve without a Redrocket card: I get halfres good at realtime and 4fps rendering at Full Res with my Redrocket turned off. It's still much better with the Rocket. I run a beefy 12core/24GBRam/Quadro400/GTX285 Macpro though...
 
No conspiracy theories here, just facts.
No later, than 3 days ago I had a long conversation with a senior product manager for a MAJOR color grading platform manufacturer. According to him, they are ready to do GPU debayer right now, but are not allowed by Red. Moreover, he can't even get one of the principals from Red to answer this very question on the future direction of the GPU debayer. So, again, according to the same senior product manager, because he can't get the answer from Red on a future GPU support, he's committing his engineering team to the GPU debayer on ARRI, SONY and Phantom. And frankly, he's not alone. I can't remember, when last time I had a Red project out of Europe. It's all Alexa now. It has nothing to do with the look, resolution, price etc. They just like it, because the post is so simple...
 
Even Adobe is not making much use of the GPU for decoding, they use it for effects. If you have a really beefy machine, you can work quite well without a Rocket.

Plus, the Rocket is giving you video out, which Adobe can't do when CUDA is active – too much shuffling around of data between the two cards.

IMHO, RED has chosen wisely and is giving you options. I can see no conspiracy here.
Yes, Adobe is in the same boat as everyone else- they're not allowed to do GPU debayer. BTW, Speedgrade, which is owned by Adobe now, successfully demo'ed GPU debayer at last NAB.
All shuffling, that is going on between the two cards- is unidirectional. RR to GPU. There is no way to feed the signal back into RR. For example, that is why you can't get the windows or secondaries correction on the output of RR.
What options do you speak of?
 
If you are working with RED files everyday then a RED ROCKET is a no brainer and you should order one today.

I transcode all of my RED stuff to Avid DNX-185x and work from that.

I rarely go back to RED files as I perform a "best light" in Red Cine X first, edit the stuff and then finish in Resolve.

Its only when the source material is poorly lit. IE overcast day etc that I conform from the RED files in Resolve to finish, and then it is also nice having the RED rocket in the resolve box also.

My resolve has 2 x 4000 cards in it via a cubix expander and that debayers half good in realtime very well without a rocket card.
 
RR on 4k & 5k no problem....but HOLY CRAP---shoot 5k HDRx and it's like i'm using a 2004 macbook pro with 2gb of ram and 4k footage....SLOOOOOW. I suspect this is the reason folks R running DUAL Rocket Cards??? I'm very careful now about shooting HDRx for the very reason that it's a killer play in RCX.
 
BTW, Speedgrade, which is owned by Adobe now, successfully demo'ed GPU debayer at last NAB.

Actually, Iridas demo'd GPU debayering - with Red footage - at least 3 years ago, as I recall at NAB 2008. Red put a stop to that, and also put a stop to Cineform's direct transcoding of Redcode Raw to Cineform Raw, demonstrated at the same show. In retrospect, I understand why Red did that at the time, citing concerns about quality at a time when there was a lot of doubt as to what they could and could not pull off. It's a different time and a different market now, and both hardware and software are a lot more sophisticated. And GPU hardware in particular is immensely powerful and very, very affordable due to its ubiquity. I tend to think that the evolution of the SDK and the attractiveness of commodity hardware are factors that didn't exist 3 years ago, and that sole control over the debayer process should probably not be the issue it was then, when Red was trying to prove to the industry and the market that it could deliver what it claimed. And while $5000 for a Rocket card is not a lot in the scheme of things, it's still a lot more than $200 for a very powerful CUDA enabled GPU. Personally, I would like to see them open up to GPU based debayering, possibly doing some sort of certification process for prospective vendors that would allow them to offer alternatives with Red's blessing. I certainly can't and won't speak for Red, but I do believe that the degree of protectionism that might have been important in the past is probably becoming a bit counterproductive in a world in which their products are already proven and viable, and in which any reliance on proprietary computer hardware is becoming less and less attractive.
 
I am so glad I bought a Redrocket card.

I'm especially looking forward to what Red will hopefully introduce with RedRayPro, or some other hardware platform that will allow me to work more quickly with .R3D footage.

I think it is Red's prerogative to protect the technology behind .R3D. Good on them.

You could either spend between $3000 and $4750 to get a pretty simple hardware solution to your problem, or if you want to be a Rebel - go ahead and hire some really good software engineers to try and reverse engineer the debayering process and all related algorithms and port it as a CUDA plugin - but I have a feeling it will cost you over 100,000 dollars in development and probably some legal headaches. Is it worth it? you tell me.
 
No conspiracy theories here, just facts.
No later, than 3 days ago I had a long conversation with a senior product manager for a MAJOR color grading platform manufacturer. According to him, they are ready to do GPU debayer right now, but are not allowed by Red. Moreover, he can't even get one of the principals from Red to answer this very question on the future direction of the GPU debayer. So, again, according to the same senior product manager, because he can't get the answer from Red on a future GPU support, he's committing his engineering team to the GPU debayer on ARRI, SONY and Phantom. And frankly, he's not alone. I can't remember, when last time I had a Red project out of Europe. It's all Alexa now. It has nothing to do with the look, resolution, price etc. They just like it, because the post is so simple...

Hi jake, which product of sony that needs debayering ? I'm not aware of one. And also, isn't ARRI performs its debayer right inside the camera itself and no further debayering needed in post ? That's why the filesize is so huge. As far as I'm aware of that's where the differences between red and arri format. Arri RAW is an already debayer image?

Rivai
 
No conspiracy theories here, just facts.
No later, than 3 days ago I had a long conversation with a senior product manager for a MAJOR color grading platform manufacturer. According to him, they are ready to do GPU debayer right now, but are not allowed by Red. Moreover, he can't even get one of the principals from Red to answer this very question on the future direction of the GPU debayer. So, again, according to the same senior product manager, because he can't get the answer from Red on a future GPU support, he's committing his engineering team to the GPU debayer on ARRI, SONY and Phantom. And frankly, he's not alone. I can't remember, when last time I had a Red project out of Europe. It's all Alexa now. It has nothing to do with the look, resolution, price etc. They just like it, because the post is so simple...

Hi jake, which product of sony that needs debayering ? I'm not aware of one. And also, isn't ARRI performs its debayer right inside the camera itself and no further debayering needed in post ? That's why the filesize is so huge. As far as I'm aware of that's where the differences between red and arri format. Arri RAW is an already debayer image?

Rivai
 
Hi jake, which product of sony that needs debayering ? I'm not aware of one. And also, isn't ARRI performs its debayer right inside the camera itself and no further debayering needed in post ? That's why the filesize is so huge. As far as I'm aware of that's where the differences between red and arri format. Arri RAW is an already debayer image?

Rivai

SONY F65.
ARRI ALEXA shoots RAW, that needs to be debayered, just like any bayer pattern camera. That is unless it records straight to Prores.
 
I am so glad I bought a Redrocket card.

I'm especially looking forward to what Red will hopefully introduce with RedRayPro, or some other hardware platform that will allow me to work more quickly with .R3D footage.

I think it is Red's prerogative to protect the technology behind .R3D. Good on them.

You could either spend between $3000 and $4750 to get a pretty simple hardware solution to your problem, or if you want to be a Rebel - go ahead and hire some really good software engineers to try and reverse engineer the debayering process and all related algorithms and port it as a CUDA plugin - but I have a feeling it will cost you over 100,000 dollars in development and probably some legal headaches. Is it worth it? you tell me.
No need to reverse engineer it, which would be illegal. Bayer pattern is widely used by many camera manufacturers, thanks to Kodak. It's the debayer algorithm that is the key. No need to hire engineers, software companies are ready to implement it, as soon as Red would allow it.
If Red feels, that seeding the market to ARRI, because ARRI's post solution is easier to use, while Red decides to make little extra money with RR sales, it's Red's prerogative.
Extra time necessary for transcoding, extra DIT charges, re-linking, reconforming, extra hardware cost... well it's all adds up...
 
According to him, they are ready to do GPU debayer right now, but are not allowed by Red....he's committing his engineering team to the GPU debayer on ARRI, SONY and Phantom.

The RR does more than just debayer. It also has to perform the wavelet decompression. Isn't Arri uncompressed RAW? If so, no decode is necessary. The F65 uses 4:1 lossy compression. I would bet decoding is more complex than the debayer. As I posted earlier, H.264 decode on a GPU is not actually done by the GPU itself. There is a piece of dedicated hw on every GPU for H.264 decoding. They are now starting to include encoding hw as well. When Vista was first released, ATI removed the decoder hw, from one of its mainstream GPUs, to cut cost. Decode performance was in the toilet and the hw was added back on the next generation of GPUs. This was MPEG-2, not even H.264. Users of that particular hw had to switch to software decoding, which was much faster.
 
seeding the market to ARRI, because ARRI's post solution is easier to use

There is nothing easier about the Alexa post solution. That is Arri marketing BS. I have done tons of Alexa jobs and Alexa is just as much work, if not more, that Red with Rocket. Alexa is direct to edit only if you shoot Prores422 Rec709 and why on earth would anyone do that?

I agree that Red should allow companies to do their own debayer with approval from Red. Kinda like Red is doing with lens mounts on Epic. Build it, show Red how it works, and they approve it or give you feed back on what they don't like. I remember seeing the demo of Red debayers a couple years ago at NAB, so we know it is possible.



Dusty
 
There is nothing easier about the Alexa post solution. That is Arri marketing BS. I have done tons of Alexa jobs and Alexa is just as much work, if not more, that Red with Rocket. Alexa is direct to edit only if you shoot Prores422 Rec709 and why on earth would anyone do that?





Dusty

Alexa records 444 12 bit Log-C Prores, which is vastly different from Red's offering of 10 bit 422 recording.
So, on the comparable simplicity of working with Prores vs. RAW and Alexa vs Red marketing BS, let's agree to disagree...
 
Jake,

I don't know what you mean by Red 10bit 422, so I am going to skip that part.

Alexa Prores 444 LogC vs Redcode.

Step 1 - Downloading
Filesizes are about the same, so download should be a wash, except there is only one fast SxS reader and it is VERY BUGGY (QIO). All other SxS readers are USB and very slow. Red is the winner here without question.

Step 2 - Prepping for Editor
Off the camera both formats need to be graded and converted to something editors can use. Most professional editors use Avid, so it needs to be in an Avid format. Prores is not an Avid format. Yes you could use AMA, but that doesn't work for most editors. They need MXF files. With Red, just load the R3D files in Redcine, do a quick grade, if you don't like what was on the camera and render to DNxHD MXF files with your Rocket. With Alexa I can AMA my footage into Avid, then transcode it to DHxHD MXF, but my editor will be working with LogC and no editor wants to deal with that. There is no good way to grade the footage in Avid and then transcode to MXF. So I have to use something else. I could load the footage into FCP, grade to taste with basic tools and render out a new graded Quicktime. The best option would be DHxHD, but FCP just doesn't seem to like DHxHD and I always have issues. So you cna render out a new graded Prores file. Either way once you get back to Avid, you still have to AMA and then transcaode the AMA to MXF. So what took me one easy render with Red and saved metadata takes two renders in FCP and Avid and does not save any metadata. I could also use Resolve and grade and render out graded MXF files that go straight into Avid, but that is a lot of tool to replace Redcine and no metadata. I could see a draw here, but I really think Red is the winner.

Step 3 - Grading
Getting from Avid to a grading system and re connecting to the original footage. This seems about the same. Working with the footage in the grading system. Yes, Prores 444 is less processor intensive. With a Red Rocket, not so much. I could see Alexa wining here, but I think it is a draw.

Please explain how Alexa Prores 444 is easier to work with than Red with Rocket? If it was just a test of if I put the same blur on each clip, which will render faster, Prores would most likely win. But when you look at the entire workflow, I see Red as the winner.



Dusty
 
SONY F65.
ARRI ALEXA shoots RAW, that needs to be debayered, just like any bayer pattern camera. That is unless it records straight to Prores.

Or unless you're recording uncompressed HD video, either single or dual link using an external recorder. The video outputs can be set to either single or dual link HD video, or T-link for RAW recording. ProRes is not the only video option.
 
I could also use Resolve and grade and render out graded MXF files that go straight into Avid, but that is a lot of tool to replace Redcine and no metadata.

A lot of tool compared to what? Resolve is $1000. A Rocket card is almost $5000. The computer to house each is essentially the same, except that you can't put a Rocket card in a laptop without an external third party device. Seems to me that Resolve is a lot less tool than your Rocket setup.


Please explain how Alexa Prores 444 is easier to work with than Red with Rocket? ... when you look at the entire workflow, I see Red as the winner.

Sorry, I have to agree with Jake on this subject. And so does essentially the entire television production industry. I'm not saying it's better. I am saying it's easier and simpler.
 
The only RED needs to do right now is finalized and released the proxy module. I thought that was the plan before and Jim posted it.

If we have to use 3rd party external recorder like AJA Ki Pro, we will stuck to HD-SDI max framerate and cannot shoot higher framerate from EPIC, we have no idea on the debayer quality as well.

So choices will be good. Beside we already has RC2 and RLF

Rivai
 
Back
Top