PDA

View Full Version : For Luki



Richard Lackey
04-21-2008, 08:18 AM
Hi Luki

Is there any noticeable difference conforming and finishing with the native .r3d files in Scratch compared to transcoding them first to uncompressed DPX first?

Thanks,

Rich

Lucas Wilson
04-21-2008, 10:40 PM
Hi Luki

Is there any noticeable difference conforming and finishing with the native .r3d files in Scratch compared to transcoding them first to uncompressed DPX first?

Thanks,

Rich

Hi Richard,

Short answer - no. SCRATCH conforms just as smoothly from native R3D as it does from DPX header metadata.

Best,

Lucas

Richard Lackey
04-22-2008, 12:03 AM
I've been getting a bit confused, should have asked Graham Austin, but he's got about as much experience with .r3d media as I have so far.

Here's my dilema, if I transcode .r3d to DPX, I'm going from a compressed format to uncompressed, however, since the media was compressed in the first place, I'm kidding myself if I think the resulting DPX's are free of any compression artifacts. In otherwords, the DPX's can't be any better than the RedcodeRAW, right?

So if I decide to transcode, I load up my 4K Redcode RAW in Redcine, set my output settings (lets say I'm going to 2K DPX) and hit the "Go!" button. Now I wait for hours (even days) and if I'm lucky I end up with what I asked for. Now I've used a good chunk of my drive array, and I can then conform and grade to my hearts content in real-time.

If I skip this whole process and conform to the Redcode RAW directly, I'm saving time transcoding, and using less disk space. I can still conform and grade real-time right? Up to 2K res?

I think my original question had more to do with the resulting image quality. Am I getting a better picture going the DPX route than the Redcode route?

I know that if I'm compositing or doing anything where more than one generation of compression is taking place, I am better off transcoding to DPX first, right?

Now, my system is not tied to a Brightdrive 4K SAN, so for me, uncompressed 4K real-time is not an option because of hardware limitations. I understand that 4K redcode is also not an option real-time, not because of drive bandwidth this time, but because Scratch won't decompress the Redcode RAW in real-time at that resolution, so that's a CPU limitation right?

One last thing. My friend Caleb wrote me an email, saying this

"Ted mentioned that by linking RAID drives (can be done through Apple Qmaster), you could get real time (1:1) 4k DPX output from Scratch."

Can you shed any light on this, it doesn't quite make sense to me, I'm not sure what he's saying. I already know Scratch will handle 4K real-time if it can pull the data quick enough, so is this a new way of striping drives? Not sure what he means by "linking" RAID drives.

I know it's a lot of questions. Sorry about that.

Richard Lackey
04-22-2008, 03:07 AM
A friend of mine just called after reading my post above. The answer to the "linking" RAID drives, seems to be adding port multipliers. For some reason I had considered this a no-go. It makes sense though, one physical drive connected to one port is never going to use up it's 3Gb/sec bandwidth on the port, so why not stripe four or five drives to the port instead?

Any implications of this with Scratch?

In theory I could add 5-way port multipliers to my 8 port 3ware controller (9550SXU... I checked it does support port multipliers) and have up to 40 drives connected. It seems the card is limited to 800MB/sec, so that's not enough for 4K, but another controller could be added instead with the drives spread out between them over 16 physical ports (2 x 8 port cards), so perhaps two drives to each port, or 32 drives in total, and that would give me 32 x 300Mbit/sec = 1200MB/sec off the drives out of a maximum 1600MB/sec out of the two controller cards.

Make any sense?

Radoslav Karapetkov
04-22-2008, 07:07 AM
Mr. Luki,


Aren`t Assimilate considering something for the future Scarlet nation as well?

As this post from darkkelt suggests? :

http://www.reduser.net/forum/showpost.php?p=186982&postcount=90

It's funny but it has a point, I think.

So,

A 3K Assimilate Itch for 3k $ ?

:)

Richard Lackey
04-22-2008, 08:45 AM
Mr. Luki,


Aren`t Assimilate considering something for the future Scarlet nation as well?

As this post from darkkelt suggests? :

http://www.reduser.net/forum/showpost.php?p=186982&postcount=90

It's funny but it has a point, I think.

So,

A 3K Assimilate Itch for 3k $ ?

:)

I think a lot of full Scratch users are probably doing the bulk of thier day to day work at 1080 or 2K, so a 2K "Itch" would upset a lot of people who have spent the big bucks on a full Scratch DI suite. For this reason I doubt anything like this would happen.

I don't think the majority of Scarlet users are going to be in the uncompressed D.I. arena to be honest.

If there was a way to separate a full Scratch with all the bells and whistles from a "prosumer" version enough that the two didn't compete at all, there could be a lot of revenue in a high volume product.

None of us are Assimilate though, it's not our business, and I think they have a great product at a very competitive price point when compared to Lustre, Baselight, DaVinci and all the others. It's out of the budget of home users or hobbyists sure, but looking at it objectively, Scratch is a great deal already. It's like complaining that the Red One isn't selling for $3000 and being blind to how much value it represents against any competing digital cinema camera.

Some things will never be sold off the shelf out of a box to a consumer market.

Gunleik Groven
04-22-2008, 08:57 AM
Many RED/EPIC users will buy Scarlet asa 2nd VFX, whatever cam and shoot it @ 100 MB/sec...

Gunleik