RED Rocket TURBO, WHEN?

[mikeburton]

MTI looks like an interesting entity. Wasn't aware of them...

Things were done toavoid alternate debayer methods to REDs own, after the Cineform controversy.
And that's when not even Scratch could do realtime full debayer anymore. Reintroduced with the rocket later on.

But unless there is a third layer (above debayering and wavelets), it is kinda hard to seewhy gpu's now shouldn't cut it.

Afaik that's exactly why we see the different render speeds in Resolve and RCX.

It would be extremely difficult to go the GPU route after all these years from a perception perspective alone. Red has preached that there is no way to do it on the GPU and the Rocket is the only alternative. And after the backlash from lowering the camera price points not long ago this could open up another can of worms if customers spent $10, $15, 20k or more in bunk hardware only for Red to make it obsolete in a day after they were told it was the only way for so long. It's easier at this point to just stick with the status quo and maintain the front. If indeed it is a front of course and although I tend to be a skeptic on this particular subject and have not been shy in other threads about my disgust of price versus lack of warranty protection and failure rates of these cards we still can't say they've been lying to us because of a back room deal they made a long time ago as Jake suggested because we don't know if that's the case. Although I do happen to subscribe to that theory as it would make a lot of sense.

 

[jake blackstone]

Red, as the owner of R3D IP can do whatever they please. They don't have to explain the reason why they doing debayer only on CPU or the RR to anyone. Said that, it's getting increasingly difficult to keep dodging the question of lack of the GPU debayer. Why for example, even, if the quality is not up to par, can't GPU debayer be allowed for editing and grading? This way one could grade in "full quality" and then only CPU could be used for render. I, for one, would love that...

 

[Mark L. Pederson]

Why for example, even, if the quality is not up to par, can't GPU debayer be allowed for editing and grading? This way one could grade in "full quality" and then only CPU could be used for render. I, for one, would love that...

Amen.

 

[Brandon J.F.]

Why for example, even, if the quality is not up to par, can't GPU debayer be allowed for editing and grading? This way one could grade in "full quality" and then only CPU could be used for render. I, for one, would love that...

Agreed.

Based on a screenshot they posted here of the .RED encoder plugin for RCP-X we will have the option to use a GPU to help encode .RED footage. Obviously, it's not clear where the GPU will help in the process, but it's promising to see...

 

[jake blackstone]

Agreed.

Based on a screenshot they posted here of the .RED encoder plugin for RCP-X we will have the option to use a GPU to help encode .RED footage. Obviously, it's not clear where the GPU will help in the process, but it's promising to see...

There is no question, that Red has compression algorithm down pat, which what encoding to .RED really is. Decompression is what questioned in this thread.

 

[Gunleik Groven]

Amen.

Agreed.

If it is not to hinder that signal to be recorded over sdi.

But red has the tech. The 7k Scarlet does realtime debayer and scaling of 5k, so that makes it somewhat cheaper to buy a scarlet as a deck, than a turbo - rocket... (Yeah, sure... I k ow it's not comparable. But it IS, as to debayering and scaling...) the tech is all there inside the cameras...

 

[Brandon J.F.]

There is no question, that Red has compression algorithm down pat, which what encoding to .RED really is. Decompression is what questioned in this thread.

Yeah, I understand that. I was thinking about the debayering that would still need to be done when converting from a .R3D to a .RED file. Unfortunately, it's not likely that debayering has changed with RRENCODER. Wishful thinking on my part...

 

[jake blackstone]

The 7k Scarlet does realtime debayer and scaling of 5k

If Red has the knowhow and capabilities to produce such a device themselves, why wouldn't they then produce something like RR at greatly reduced price? Red is a commercial enterprise and the reason for such an enterprise is to meet it's customers needs and to make money for the enterprise. For the last three years or so, when it comes to post, Red did neither. Doesn't it makes you question Red's motive?

 

[Gunleik Groven]

I am not questioning anything.
Just saying what happens when you hit play on a Scarlet.

That box has some additional tricks, though... Like a sensor an OLPF, an AD and a compression engine + the capacity to encode r3d's from linear light raw.

 

[Nikhil Kamkolkar]

Those arguing for a cheaper Rocket are missing the point. The proprietary card approach needs to be rethought and replaced if wider acceptance is the goal.

Agreed to the nth degree.

This thread is actually getting me quite upset. Some of the theories really are making me feel like RED is not an indie filmmaker's friend, as I have been imagining. (

A response is necessary. A roadmap for post is necessary. Trust is necessary.

 

[Gunleik Groven]

Agreed to the nth degree.

This thread is actually getting me quite upset. Some of the theories really are making me feel like RED is not an indie filmmaker's friend, as I have been imagining. (

A response is necessary. A roadmap for post is necessary. Trust is necessary.

Actually... THAT's just silly. RED has adopted current tech to make workflows achievable at any time.
And if RED hasn't been a blessing of the indies, it's hard to imagine who should feel blessed.

What is pointed out is that tech and competition has evolved a bit since fall 2009...

And that that might be a thing to give 2nd thought...

 

[Bill Anderson]

As far as what RED has "preached", they have also repeatedly admitted that they are a small company, figuring things out as they go. That might be a bit of an overstatement when one considers just how much red has achieved compared to the "big boys", but RED is in the clear if it decides to change its "post" course (or any other) radically. Surely this is how companies ( and filmmakers) compete and thrive….

There is a reason it is called the "cutting edge".

 

[Steve Sherrick]

Since it's in Red's best interest to make their R3D files as easy to deal with in post as possible, it seems to me there must be more to this than has been revealed to this point. I do believe the RR is a 3rd party card (DVS likely) and that would make a ton of sense for Red to go that route. I'm having a hard time wrapping my head around the idea that they are locked into a longterm deal though. With the way technology moves along this would seem like a short sighted strategy. But perhaps that was what happened. Without confirmation from Red, there is no way to know.

In terms of GPU being useful for R3D processing, I wonder if there is some special sauce in the compression that is not able to play nice with GPUs. Unfortunately there are no whitepapers on this stuff and no way to know what's actually going on that prevents harnessing the full power of GPUs in the processing of R3Ds.

Right now, I'd be in a tough position without a Rocket. The workflow is heavily tied to that proprietary device which has been okay but I think for the wider acceptance in post, an alternative that relies on commodity, off the shelf hardware is in order. If it is not technically possible to do that, perhaps a clear explanation (RED Paper) would go a long way to helping the post community understand the reasoning. Right now, the perception is that Red workflow is cumbersome unless you are a DIT or Red-centric post house. An accessable solution, if possible, would get the post people off of producer's backs and allow them to make decisions based on production's needs.

2013 is probably the year to make a statement in this regard.

 

[Nikhil Kamkolkar]

Actually... THAT's just silly. RED has adopted current tech to make workflows achievable at any time.
And if RED hasn't been a blessing of the indies, it's hard to imagine who should feel blessed.

What is pointed out is that tech and competition has evolved a bit since fall 2009...

And that that might be a thing to give 2nd thought...

I think we agree (despite your characterization of my thoughts as being silly). All I've heard from RED recently is that RR is the way to go today and for the forseeable future. I've heard there'll be an upgrade to it, but it'll only be way more expensive and NOT indie-focused whenever it arrives.

Scour the forums for the current RR reliability/warranty issues.

So... where does that leave post? Is there a second thought being given to all this by RED? Enlighten me if I'm missing something here.

I'm ofcourse happy with my EPIC. I want to be happy about the entire workflow. Surely, that is not an unreasonable or silly stance, is it? I'm hoping NAB reveals pleasant surprises. They just have to be very pleasant at this point to count given how far the competition has come along. This thread is an opportunity to let RED know this is important to some of us. I'm sure the other things they are working on are important to others. Post workflow needs to get the attention its due is all.

 

[Gunleik Groven]

Sorry! No personal insult intended!

 

[Gunleik Groven]

Point is more like, initially there was GPU accelleration available, at least through Scratch.
But Assimilate possibly had a proprietary debayering going on early on.

Then Cineform came up with their solution
(which as far as memory serves was realtime, possibly GPU accellerated, and Graeme demonstrated issues with that piece of code)

So Cine forme released a new debayer routine, and the proprietary debayering option was shut out.
Gone was realtime as well

SDK ruled
And a year later the RR was introduced at IBC in a joint presentation with Assimilate (this was still before RC-X
Scratch was once more realtime.
But this time with Rocket and not a PNY QFX 4000 (with SDI breakout)

Since then, the data/pixel-load has increased quite a bit, while the Rocket is still (according to RED) the same.

Now...

R3D is the diamond in REDs chest, which Jim has pointed out a number of times.
Messing with that, gives reason for severe bites from RED - understandably...

Along the way, thougfh, RED has demonstrated different full debayer/realtime solutions for the format.

1st was Scratch (GPU)
2nd was DVS Clipster
3rd Was the initial RedRay (if memory serves me)
4th was Scratch + Rocket
5th was Epic
6th was Scarlet

So the tech is there, but it is possibly hard to pinch on a board.

Or maybe THAT board is the TURBO

As to the "silly" remark, it was meant like:
RED has a history of making current tech available at a comparably good price when they find solutions, and they are (rightfully so) very loyal to their partners and the solutions they come up with at critical times.

Without implying any parallells, it was good for RED when they could finaly release their own RC-X with conform capacity, though.
Even though we all cherished Crimson and Ian Bloom to make stuff work for us, back in the day.

It's kinda more convinient to be able to conform a full project and not just a couple of minutes, without Scratch...

Gee, the world of conform has changed a bit since 2008... Come to think of it!

 

[jake blackstone]

In terms of GPU being useful for R3D processing, I wonder if there is some special sauce in the compression that is not able to play nice with GPUs. Unfortunately there are no whitepapers on this stuff and no way to know what's actually going on that prevents harnessing the full power of GPUs in the processing of R3Ds.

Except Assimilate and MTI definitively demonstrated, that GPU debayer works or should I say, worked. That is until they received the phone call from Red and were told to knock it off.

 

[Björn Benckert]

We got a rocket / laptop / flight case thing to bring to shoots. That we use for (most of our) clients that have not moved on to Premere offline editing. But it's quite clear that for the most part a rocket is actually not needed. But for large post houses and such that actually need the acceleration for long format / grading or such then the price tag on the rocket is pocket change. Think of the fact that a standard priced colorist charge more than the price for a rocket a day... So if you want/need speed then buy a few and fill up your render stations with them . Having a few Red Rockets is definitely the most bang for buck way to handle 4k+ raw files.

Think of all the guys that bought Codec recorders and such they need to move a totally different amount of data, have no acceleration available and with the announced Alexa XT their investment is very soon to be worth close to nothing. The current red rocket cards on the other hand will most likely still be able to process 6k (even if doing so at a slower rate). And having 4k playback from 5k or 6k files is not really needed in a grading situation so playing back at half res is quite fine and then render your master files from the same file at 4k is still a quite an outstanding way of doing things. The options are not many and does not come even close to the price.

 

[Gunleik Groven]

Except Assimilate and MTI definitively demonstrated, that GPU debayer works or should I say, worked. That is until they received the phone call from Red and were told to knock it off.

& Cineform... (I know a couple of others, too, but they didn't do it publicly)

 

[Steve Sherrick]

Except Assimilate and MTI definitively demonstrated, that GPU debayer works or should I say, worked. That is until they received the phone call from Red and were told to knock it off.

Assuming this is true, and I do know companies were experimenting with GPU early on, the only benefit for RED would be:
1. Protecting image quality.
2. Protecting a partnership deal, IF there is one in place.

The downsides would be:
1. Making hardware acceleration proprietary via RED Rocket
2. Forcing those without RR to have to scale quality based on CPU power and their turnaround times, not ideal for post houses
3. Creating a perception that RED workflow is cumbersome due to needing a RR to transcode full res media in realtime
4. Cost of investment. At full retail the card is $4750. Battle Tested (when available) is $3000. A post house that works on mission critical projects will need at least two of them. $9500. A very limited warranty. Now, just a few years ago $9500 was not considered expensive in post. Tape decks would run you $10-70K, so if you consider Red Rocket in that context, it would not be prohibitively expensive. But everything has changed. We are now in a file based world. The cost of NLEs has dropped dramatically, and so have the other tools. Look at Davinci Resolve. The Lite version is a nearly full featured tool for free. Bump up to the full featured product for $1K (which is a no-brainer) and we can see how much has really happened since the late 2000's. With boutique shops popping up all over the place and even one-man-band shops getting in the mix as well investing in heavy iron is trending towards the minority it seems. Not to mention the shakeup that Red themselves made with the introduction of the Red One. So, given this current landscape and the fact that RED cameras are owned by big rental companies, small rental companies, production companies both big and small, as well as DPs/DITs/ACs, and even post houses it would seem that if a GPU solution was feasible that it would make a LOT of sense for RED to do it because think of the impact. Let's say you have a shoot with an Alexa, Epic, and BMCC. Processing dailies through something like Resolve would be straight forward. The GPUs would handle any kind of processing needed and you could have a hardware configuration that works for multiple formats rather than being compromised because of one format.

There has to be more to this than is being told. I'm not ready to accuse RED of limiting our options. I will admit though, there are some compelling theories out there. So until RED addresses this, I will continue to use the tools that are available to handle R3D. Hopefully at some point there will be more options available. Until then, I am treating RR like a princess and hoping she'll always love me back.