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 :)

RED Rocket TURBO, WHEN?

From Lucas Wilson back in 2008.
Lucas Wilson said:
REDCINE and SCRATCH are GPU accelerated today. But the debayering is not done on the GPU. It's done on the CPU.

There are many, many ways to debayer and demosaic an image. Unless there is dedicated hardware involved, I seriously doubt they are doing a full debayer/demosaic from 4K in realtime.

Also, for debayering/demosaicing, GPU acceleration won't help much. All that work is typically done on the CPU. GPU threads are great for color grading, which is why when you adjust a control on REDCINE and/or SCRATCH, the feedback is instantaneous.

Lucas
-----
ASSIMILATE, inc.
LA, CA, USA
Click to expand...
 
From Graeme
Graeme Nattress said:
Ronx, "robust" is the key word there :-)
I've had demosaic running on GPU since before RED started, but the current CPU one is better. Also, pulling 4k data back off the GPU is still a bottleneck.

Graeme
Click to expand...
 
From Deanan
Deanan said:
Agreed. I had a similar 4k debayer to 4k monitor working at another unnamed company I used to work at about 2 years ago. To get realtime you are limited in the number of operations you can do and that limits the debayer quality. Beyond a somewhat simple debayer you start to lose the benefits once you factor in texture uploads and downloads and an increasing number of memory accesses and branches which aren't super fast on gpus like matrix multiplies are.

Also, going from uncompressed to gpu is alot different than decompressing and then going to gpu.

On top of that, supporting more than a couple of graphics cards is a nightmare as no two models really function the same.
Click to expand...
 
Björn Benckert said:
Think of the fact that a standard priced colorist charge more than the price for a rocket a day...
Click to expand...
Standard priced colorist charges more than $5k a day? May be 10 years ago working in multimillion dollar telecine suites or at CO3 now, but I'm not aware of any regular colorist, who charges anything even close to that for a day. May be a week...
 
Steve Sherrick said:
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.
Click to expand...
3. Additional income from resale of the card.

Steve Sherrick said:
The downsides would be:
3. Creating a perception that RED workflow is cumbersome due to needing a RR to transcode full res media in realtime
Click to expand...

A mighty big one I'd say.
 
Björn Benckert said:
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.
Click to expand...

Bjorn, regardless of the fact that the price of the card may or may not be pocket change for post facilities the remaining fact is that its cumbersome to need a card that only targets one codec, is flakey at best when it comes to reliability and has a terrible warranty for something so expensive. Think if you needed a $5,000 card for all sony cameras ie F65/F55/F5, one for Arri, one for Cannon, one for Aaton, one for BMCC, Panasonic, Phantom, SI, etc etc etc. Sounds crazy, right? BECAUSE IT IS!!! I don't care if you are C03, Deluxe/EFilm, Technicolor, or any of the other large post facilities, the last thing you want to invest in is a card that only holds one purpose and has a very high failure rate which needs to be replaced all the time.
IMO, This is an issue by many levels of progression within the post industry and letting the camera manufacturers know that this is NOT an acceptable means of dealing with any codec moving forward from any manufacturer. I would like to think that you would want the same for all camera manufacturers moving into the future of their development. Because i certainly don't have the PCI slots to accomodate even half of the camera manufacturers listed above.
 
Steve Sherrick said:
From Lucas Wilson back in 2008.
Click to expand...

In terms of technical advancement, a statement from 2008 has about as much validity as one made in 1988. The differences between video/GPU cards from 5 years ago and those being made today is vast, as is the power of today's CPU's. There is also the undeniable reality of what everyone else seems capable of doing. Statements that deny the ability of commodity hardware to accomplish these tasks flies in the face of all common evidence, regardless of what might or might not have been the case 5 years ago.
 
M Most said:
In terms of technical advancement, a statement from 2008 has about as much validity as one made in 1988. The differences between video/GPU cards from 5 years ago and those being made today is vast, as is the power of today's CPU's. There is also the undeniable reality of what everyone else seems capable of doing. Statements that deny the ability of commodity hardware to accomplish these tasks flies in the face of all common evidence, regardless of what might or might not have been the case 5 years ago.
Click to expand...

This will indeed be a bit interesting.

But maybe all the others suck? :)
 
Even though Red is a small company, it's quite sad that they've not been able to give us a better realtime solution for their r3d's. Rocket would be okay if it could deliver 5k with some headroom. But from what I've been told it's not possible with 5k on the Rocket in real time?
 
M Most said:
In terms of technical advancement, a statement from 2008 has about as much validity as one made in 1988. The differences between video/GPU cards from 5 years ago and those being made today is vast, as is the power of today's CPU's. There is also the undeniable reality of what everyone else seems capable of doing. Statements that deny the ability of commodity hardware to accomplish these tasks flies in the face of all common evidence, regardless of what might or might not have been the case 5 years ago.
Click to expand...
Mike, this was in response to what Gunleik was talking about earlier in terms of what Assimilate was doing early on, not a statement about the current abilities. There are only a few people that can say with authority what is capable with R3Ds. Everything else is conjecture right?Until there is an official solution other than RR that can debayer R3D in realtime, I have nothing to go on. I don't deny the plausibility of some of the theories put out there, but until Red addresses this with a thorough explanation I must go on what is reality at the moment. Powerful computers do a good job. Rocket does it the most efficiently.
 
Bottomline : For RED to tout REDCCODE, the R3D container, and the RAW workflow, and then turn right around and chop its own legs off by not providing a post workflow that is easy as pie in the diverse post world we occupy, that has GOT to be some sort of a mistake.

Or maybe not. Time will tell.

All I know is, as a matter of principle, based on the cost, the age, the lack of a solid warranty, and the flaky reputation the RR has, I'm never buying a RR. Now, its off to figure out what to do next given that constraint. The future doesn't look very bright or comfortable.

Maybe NAB will change all that. One can only hope.
 
Nikhil Kamkolkar said:
Bottomline : For RED to tout REDCCODE, the R3D container, and the RAW workflow, and then turn right around and chop its own legs off by not providing a post workflow that is easy as pie in the diverse post world we occupy, that has GOT to be some sort of a mistake.

Or maybe not. Time will tell.

All I know is, as a matter of principle, based on the cost, the age, the lack of a solid warranty, and the flaky reputation the RR has, I'm never buying a RR. Now, its off to figure out what to do next given that constraint. The future doesn't look very bright or comfortable.

Maybe NAB will change all that. One can only hope.
Click to expand...
Okay, let's all take a deep breath. Remember, not everyone is suffering post issues. There are a lot of different workflows, some work very well for people. Other camera formats have their pros and cons as well. It's all how you build the pipeline to deal with them. I understand that the Red Rocket is not ideal for everyone right now. I get that. However, it's not the only way to handle Red files. So, while the issues being raised here are valid and could maybe use some additional clarification from RED the bottom line is that people have found ways to make the RED workflow easy for them, or at the very least a viable solution that gets the job done. There have been Red Rockets deployed in workflows that have not had the issues that we see mentioned. Doesn't excuse the problems that have occurred with the other troublesome boards but we should try to keep everything in perspective. Money tends to solve a lot of "issues". If you have kickass computer hardware with multiple rockets (and backups) and a good software toolset chances are you are finding RED workflow not so bad. If you have old hardware, no available slots in your computer, no expansion chassis, no Rockets and work with many different formats chances are you are less happy about the RED workflow.

So, the question is and I guess I'll be the one to put it bluntly - Does RED have plans for GPU debayering beyond the Rocket way of doing it or is the Rocket the only way to do it with the speed, RED approved IQ, and compatibility that RED finds suitable? If it's possible to do it in GPU but the results vary too much across multiple card models, vendors, etc is there a way to do a RED certified configuration so that they can do what Blackmagic does and say these are the configurations that will get consistent results? If GPU is still not possible, is there an explanation for it? I know in the past Graeme and Jarred have stated that GPU doesn't always get the math right. Is that still the case? Is there any commodity hardware on the horizon that may change RED's approach to this? I have to admit, a $10,000 card that does two or three times realtime would be worth it to some but might alienate others even more. I see a slippery slope with that one. But that is the last official word we really have about where things are heading. So again, dealing with the reality of the moment we have the current workflow solutions available, RR being one of the more efficient ones. The Turbo Rocket seems to be on the agenda but no further details. So, where do we stand? Get stuff done with what we have, which is what I choose to do and hope some more clarification is coming about other solutions or lack there of. Or choose to not use the most efficient ways of handling Red footage and dealing with the headaches that can bring. Either way, there are choices right now. They may not be ideal (I admit, I'm hungry for PCIe slots still) but they are there. And sometimes getting stuff done is more important than trying to find the perfect solution/workflow.
 
Oh, and just to clarify I'm not saying that throwing money at the problem is the best way to deal with this. I'm in the category of not being able to throw endless amounts of money at problems. It's just one of life's cruel truths - money does tend to fix a lot. That was my only point.

I am still in the camp that would love as many choices as possible as long as they make sense. I'm not saying RR is the final answer. It's simply an answer for right now until we know where RED is heading.
 
Money. Its a gas. Grab that cash with both hands and make a stash. New car, caviar, four star daydream, think I'll buy a Red Rocket and scream...

"This is yesterday's solution. RED needs a revolution in post."

I for one, have been patiently waiting for it. And am severely disappointed that all the noises coming out of RED suggest easy, no-nonsense Post is not a priority for them.
 
I find it interesting, when Red loves to repeat the story- we didn't know what we're doing, they told us we were fraud, they said it was impossible to do at that price, but we persevered and created Red Revolution. Great!
On the other hand, when it comes to post, the same people are saying- GPU debayer is too difficult, GPU debayer is impossible, GPU debayer is inconsistent, there is nothing wrong with our way of doing post, we can't do it inexpensively (Meizler module, RR Turbo), you don't need Prores recording in camera, how about H264 instead. Did I miss anything else?
 
I think much can be explained through seeing it through the glasses of "4k or nuthing" delivery.

Even though the system is well suited for other deliveryoptions than 4k.

RED is on a mission...
 
hi Will, do you any idea of who can help me with my rocket in san francisco area?
thanks
 
You must log in or register to reply here.
Back
Top