Gunleik Groven
08-24-2008, 06:56 AM
After spending some time with Scratch, I can subscribe to the choirus: Scratch is really a very good tool.
But form my uses, I would like to choose how to configure the box, because the "standard and really really good" system configs I am currently choosing from, are not really suiting my needs.
First some analysis: What makes Scratch great?
Answer:
Workflow for projects longer than 30 seconds...:) GIVEN that you do NOT have to transcode to DPX OR copy terrabytes of media for let's say a TV series or a feature.
Where Scratch really rules the ground, is where you can playout to tape at the end instead of rendering to dpx/tiff/jpeg2000 for delivery.
At once you start transcoding to DPXs for online, there are plenty of good tools out there. Scratch IS a really good with DPXs, but so are a bunch of others.
Because of how the RAW files needs to be treated to avoid noise issues, any Scratch solution without scaffolds is really not very interesting (to me).
What I need is a box that interfaces well with my and my customers storagesolutions so that one can work directly off the RAW files after FCP or AVID edit. Import RAW media (where they are located), import EDL's, conform and start with the DI process. On the other end I need a set of luts, based on a filmout lut, with one "master" for either film, DCI or REC 709, and corresponding luts for the other deliveryformats when things are corrected to the main spec.
In my case that will more often be DCI than not.
Where Scratch REALLY excells, is in the scenarios where you can have as much material as the client wants online and no interrmediate DPX/tiff/whatever file transcoding untill delivery. THIS is where Scratch is mucho sexy and is the only real option as of now.
So what are my Hardware needs?
A quick stationary box based on a Windows server OS, that interfaces well with a multitude of storage solutions and the fastest possible Quadro with SDI.
The bandwith issue to storage is NOT like with the DPX's. RED RAW currently maxes out @ 36 MB/sec. with EPIC/SCARLET it will max out @ 100 MB/sec. Actually I could do that with FW 400 disks (just as an example) I DO NOT need a 16 disk internal RAID to work with these files, and actually and truely: It is only a hazzle. I really just need a RAID1 systemdisk and good connectors to the outer world in a box that is carryable.
I dared say a MacPro box, and got myself spammed for it. It doesn't need to be, but so far my MacPros are higher spec'ed on CPU cycles, and RAM and interconnectivity than any of the Scratch systems I've used.
Then I need dual 4GB FibreChannel, eSATA and as many 1GB ethernet ports as I can eat. I need software so windows can read HFS and I need the box to have a carryable formfactor and weigh less than 30 kg.
This way I can connect the box to mine and my customers storage systems, take out ALL the workflow advantages of Scratch over other systems and make the life easier for me and my customers.
Makes sense?
In addition to this, I'd like to see an "prefferd" partner for assimilate for DCP delivieries, so that there could be a consistent deliverable workflow for DCP packages, AND a render node system, so one could use off-the-shelf boxes for file-encoding and transcoding without locking up the main system for "slave-labour".
This is the Scratch I want.
But is it deliverable?
Gunleik
But form my uses, I would like to choose how to configure the box, because the "standard and really really good" system configs I am currently choosing from, are not really suiting my needs.
First some analysis: What makes Scratch great?
Answer:
Workflow for projects longer than 30 seconds...:) GIVEN that you do NOT have to transcode to DPX OR copy terrabytes of media for let's say a TV series or a feature.
Where Scratch really rules the ground, is where you can playout to tape at the end instead of rendering to dpx/tiff/jpeg2000 for delivery.
At once you start transcoding to DPXs for online, there are plenty of good tools out there. Scratch IS a really good with DPXs, but so are a bunch of others.
Because of how the RAW files needs to be treated to avoid noise issues, any Scratch solution without scaffolds is really not very interesting (to me).
What I need is a box that interfaces well with my and my customers storagesolutions so that one can work directly off the RAW files after FCP or AVID edit. Import RAW media (where they are located), import EDL's, conform and start with the DI process. On the other end I need a set of luts, based on a filmout lut, with one "master" for either film, DCI or REC 709, and corresponding luts for the other deliveryformats when things are corrected to the main spec.
In my case that will more often be DCI than not.
Where Scratch REALLY excells, is in the scenarios where you can have as much material as the client wants online and no interrmediate DPX/tiff/whatever file transcoding untill delivery. THIS is where Scratch is mucho sexy and is the only real option as of now.
So what are my Hardware needs?
A quick stationary box based on a Windows server OS, that interfaces well with a multitude of storage solutions and the fastest possible Quadro with SDI.
The bandwith issue to storage is NOT like with the DPX's. RED RAW currently maxes out @ 36 MB/sec. with EPIC/SCARLET it will max out @ 100 MB/sec. Actually I could do that with FW 400 disks (just as an example) I DO NOT need a 16 disk internal RAID to work with these files, and actually and truely: It is only a hazzle. I really just need a RAID1 systemdisk and good connectors to the outer world in a box that is carryable.
I dared say a MacPro box, and got myself spammed for it. It doesn't need to be, but so far my MacPros are higher spec'ed on CPU cycles, and RAM and interconnectivity than any of the Scratch systems I've used.
Then I need dual 4GB FibreChannel, eSATA and as many 1GB ethernet ports as I can eat. I need software so windows can read HFS and I need the box to have a carryable formfactor and weigh less than 30 kg.
This way I can connect the box to mine and my customers storage systems, take out ALL the workflow advantages of Scratch over other systems and make the life easier for me and my customers.
Makes sense?
In addition to this, I'd like to see an "prefferd" partner for assimilate for DCP delivieries, so that there could be a consistent deliverable workflow for DCP packages, AND a render node system, so one could use off-the-shelf boxes for file-encoding and transcoding without locking up the main system for "slave-labour".
This is the Scratch I want.
But is it deliverable?
Gunleik