Recording RAW to SSD

Started by PeteTomov, June 05, 2013, 01:32:21 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Michael Zöller

One thing to think about is - would it technically work with any CF camera/equipment? That would potentially increase the audience.
neoluxx.de
EOS 5D Mark II | EOS 600D | EF 24-70mm f/2.8 | Tascam DR-40

PeteTomov

Quote from: squig on June 13, 2013, 12:18:50 PM


EDIT: 6000mAh for 2 hours? Really? The Ninja can run longer with less power.

6000mAh ~2 hours at max SDD speed. At ~90MB/s the battery should last a lot longer, probably at least twice that. Also the SSD will be powered on and off by the camera so it shouldn't draw any power when not recording.

PeteTomov

Quote from: Michael Zöller on June 13, 2013, 12:43:59 PM
One thing to think about is - would it technically work with any CF camera/equipment? That would potentially increase the audience.
Yes it will, it will act 100% like a fast CF card. The 5D should be able to do 5-6fps in photo mode without running out of buffer with a fast SSD. I'm not particularly sure if anyone needs that but I was planning on mentioning it in the kickstarter project. Somebody might come up with another use.

albert-e

Quote from: PeteTomov on June 13, 2013, 01:29:55 PM
Yes it will, it will act 100% like a fast CF card. The 5D should be able to do 5-6fps in photo mode without running out of buffer with a fast SSD. I'm not particularly sure if anyone needs that but I was planning on mentioning it in the kickstarter project. Somebody might come up with another use.

We would like to see proof of concept: A CF-SSD adapter plugged into camera and functions. A Canon 5D has the software interface if not mistaken, to work with CF Type II (such as micro drive), so it should work. Pictures (even a video) makes a thousand words.

On the same subject: http://www.magiclantern.fm/forum/index.php?topic=5470.250

Walter Schulz

Quote from: albert-e on June 13, 2013, 04:07:39 PM
A Canon 5D has the software interface if not mistaken, to work with CF Type II (such as micro drive)

This doesn't make much sense! Read http://en.wikipedia.org/wiki/CompactFlash#Type_I_and_Type_II

Ciao, Walter

AnotherDave

I'm a little surprised that we haven't seen a proof-of-concept video yet...

Malcolm Debono

Very interesting project, and sounds like you're moving in the right direction. However, I would make sure that it actually works (using cameras like the 5D) first rather than basing everything on theory. Many others have tried similar stuff which didn't work.
Wedding & event cinematographer
C100 & 6D shooter
New here?  Check out the FAQs here!


vertigopix

Quote from: AnotherDave on June 13, 2013, 04:21:37 PM
I'm a little surprised that we haven't seen a proof-of-concept video yet...

Me too...

AnotherDave

I'm starting to believe that the cross-over from CF to SATA isn't practical.  In theory, yes, SSD can record at a very fast rate.  But you'd always be limited by the speed of the interface. 

Wouldn't it be more practical (but less cost effective) to construct a CF to CF-RAID to increase the speed?  CF is an IDE interface... there have been IDE raids in the past.  Seems like it'd be the easiest way to go... also, it'd be lighter than a SSD and use less power (probably).

pascalc

Quote from: PeteTomov on June 13, 2013, 01:22:33 PM
6000mAh ~2 hours at max SDD speed. At ~90MB/s the battery should last a lot longer, probably at least twice that. Also the SSD will be powered on and off by the camera so it shouldn't draw any power when not recording.

A bit strange.
@ 5v, a 6000mAh batterie gives you 30w during one hour.
As the writing consumption of most of the  SSD is around 3w, it's not 2 but 10 hours that this battery will last.
The converter borad might consume a little bit I guess but it's not significant and in the other hand the SSD won't write continuously.
Your manufacturer didn't advised you that you can save load and space at the battery level ?

Need to see serious proofs before investing money here.


albert-e

Quote from: AnotherDave on June 13, 2013, 07:06:13 PM
I'm starting to believe that the cross-over from CF to SATA isn't practical.  In theory, yes, SSD can record at a very fast rate.  But you'd always be limited by the speed of the interface. 

Wouldn't it be more practical (but less cost effective) to construct a CF to CF-RAID to increase the speed?  CF is an IDE interface... there have been IDE raids in the past.  Seems like it'd be the easiest way to go... also, it'd be lighter than a SSD and use less power (probably).

RED cameras have SSD's...just to let you know. It's  not new to use SATA SSD's. Maximum transfer is >200.MB/s,enough to bite (byte) the dust!

Just get on with your project, it's worth it.

AnotherDave

Quote from: albert-e on June 13, 2013, 08:55:59 PM
RED cameras have SSD's...just to let you know. It's  not new to use SATA SSD's. Maximum transfer is >200.MB/s,enough to bite (byte) the dust!

Just get on with your project, it's worth it.

Yes, but REDs don't use CF to SSD adapters.  I don't believe you'll get the maximum data rate from a CF (167MB/s) to a SSD.

Steven

The right adapter will bring the right performance for the job. You just have to find the right adapter...

rlasson

My hunch is that the data rate out of the 5DIII will never reach over 167MB/sec as that's what UDMA 7 is rated at. You'd never expect a motherboard with 1.5Gb SATA connection to go over 150MB/sec, or a 3Gb SATA connection to go over 300MB/sec. You'd need a new motherboard to go over those speeds.

Still, if you could reliably record something close to 167MB/sec, that would certain help. Even just having storage capacity go higher than 128GB and also become more affordable would also help and make a CF to SSD adapter marketable.

I don't expect you to show us all the parts you're using or even picture of how you're currently using it, but right now you're just talk. Show us a screen grab of the benchmarks from the camera or upload a raw stream that breaks 120MB/sec and then you'll have our attention. You'll also likely get funded very, very quickly.

Russ

albert-e

Quote from: rlasson on June 13, 2013, 11:07:36 PM
My hunch is that the data rate out of the 5DIII will never reach over 167MB/sec as that's what UDMA 7 is rated at. You'd never expect a motherboard with 1.5Gb SATA connection to go over 150MB/sec, or a 3Gb SATA connection to go over 300MB/sec. You'd need a new motherboard to go over those speeds.

Still, if you could reliably record something close to 167MB/sec, that would certain help. Even just having storage capacity go higher than 128GB and also become more affordable would also help and make a CF to SSD adapter marketable.

I don't expect you to show us all the parts you're using or even picture of how you're currently using it, but right now you're just talk. Show us a screen grab of the benchmarks from the camera or upload a raw stream that breaks 120MB/sec and then you'll have our attention. You'll also likely get funded very, very quickly.

Russ

Maximum Transfer Rate Of Early & Late ATA/ATAPI:

Ata-4 Ultra DMA 0,1,2 aka UDMA/33
Ata-5 Ultra DMA 3,4 aka UDMA/66
Ata-6 UDMA 5 aka UDMA 100 48-bit addressing
Ata-7 UDMA 6 aka UDMA/133 SATA
Ata-8 UDMA 7 hybrid drive

Maximum Transfer Rates: Ultra DMA ATA 4-7, 66.7,100,133,167MB/s

The architecture between CFA and ATA devices are pretty similar.

A CF device in TrueIDE mode is exactly identical to any ATA/ATAPI-x hard
disk drive device. In fact some CF devices really are hard disk drives
(and not flash memory devices that look like a hard disk drive). A CF
device has all the same ATA registers, the same ATA command protocols and
mostly the same ATA command set (R/W SECTORS/MULTIPLE/DMA, SET FEATURES,IDENTIFY, etc).

...

albert-e

Quote from: AnotherDave on June 13, 2013, 10:36:12 PM
Yes, but REDs don't use CF to SSD adapters.  I don't believe you'll get the maximum data rate from a CF (167MB/s) to a SSD.


RED uses SSD's,I didn't say they use CF-to-SATA SSD's. It is the interface were are talking about: The goal is to:
1. Replace the CF with the ATA or SATA drives, meaning the CF external adapter connected to a ATA drive.The CF-ATA adapter is just a test tool to prove the concept. Red camera's and high end cameras are using the technology. It's been done, it's nothing new.

The CF and ATA/ATAPI specifications are out there, read it. And I thought you knew what you are talking about the "Kickstarter" project.

...

AnotherDave

Quote from: albert-e on June 13, 2013, 11:49:13 PM
Red camera's and high end cameras are using the technology. It's been done, it's nothing new.
...

Dude, what I am saying is that I think you're going to lose so much in the conversion to SSD that the data rate isn't going to improve or that it isn't going to be significant enough to use to record raw.

The Red uses a proprietary SSD, BMCC uses SSDs... but SSDs aren't the issue.  It is converting the CF to an SSD which is what I am saying is probably impossible or at very least going to require a lot to do it.

5D3shooter

Even if it is capped at 160MB/S won't that get us close to hitting 4K?  1920x1080 only requires ~80MB/S
Then if Magic Lantern team can optimize the code even more.  That might be just enough speed.

We're all setting ourselves up to not be able to live without RAW video, so I hope it works. 

Kickstart a motherboard mod lol, anything to not have to buy a new cam.

http://www.bhphotovideo.com/bnh/controller/home?O=&sku=964119&is=REG&Q=&A=details
Those sons of *******...lol

albert-e

Quote from: AnotherDave on June 13, 2013, 11:54:10 PM
Dude, what I am saying is that I think you're going to lose so much in the conversion to SSD that the data rate isn't going to improve or that it isn't going to be significant enough to use to record raw.

The Red uses a proprietary SSD, BMCC uses SSDs... but SSDs aren't the issue.  It is converting the CF to an SSD which is what I am saying is probably impossible or at very least going to require a lot to do it.

"It is converting the CF to an SSD which is what I am saying is probably impossible or at very least going to require a lot to do it."
How did you know that? And that is why we wish you success in your goal. Your dreams are ours too in this matter, dude.

rlasson

Quote from: 5D3shooter on June 14, 2013, 12:09:02 AM
Even if it is capped at 160MB/S won't that get us close to hitting 4K?  1920x1080 only requires ~80MB/S
Then if Magic Lantern team can optimize the code even more.  That might be just enough speed.

I like your optimism! Would be really great if they could figure out a way to compress the RAW files while it's in the buffer and then write it out. As it stands, 80MB/sec for 1080P would be 320MB/sec for 4K. 19GB/minute. 1.1TB/hour. That's a lot of data.

Russ

squig

Quote from: AnotherDave on June 13, 2013, 10:36:12 PM
Yes, but REDs don't use CF to SSD adapters.  I don't believe you'll get the maximum data rate from a CF (167MB/s) to a SSD.

I have an expresscard CFreader and I can write over 150MB/s to an SSD in my macbook.

albert-e

Quote from: squig on June 14, 2013, 12:37:12 AM
I have an expresscard CFreader and I can write over 150MB/s to an SSD in my macbook.

There's your proof. "I've been there, done that"

vertigopix


PeteTomov

Wow, I should really check the forum more often  ::).

To clarify the current situation:

The adapter has not been tested in camera. It has only been tested hardwired into an USB 3.0 reader. The version we put together over a weekend in simply not small enough to fit inside a camera and we're not looking at enough profit to justify taking apart my 5D for testing.

The main problem is the precision required when working with such high frequencies. You need machines to manufacture the components, you can't just solder some wires.I'm guessing that's why the other projects fail since everything else is pretty simple and based on existing technology.

The reason we are 100% sure the adapter will work in camera is the way CF cards work in general. They are basically faster IDE drives and can work with an IDE to SATA adapter although I haven't seen one small enough or fast enough since IDE drives back in the day were much slower. We're simply adding a processor fast enough for SATA3 speeds,batteries for the SSD and packing the actual adapter into a small enough size to fit inside a CF card slot...also the processor is fast enough to emulate anything it has to in case we're wrong  ::).

As I've mentioned earlier,we're not really doing this for profit(we're looking at about 5-10%), we both have other jobs. We're doing it because we want one ourselves. We understand if don't want to take that chance and in that case we simply won't even bother with a kickstarter. Hopefully someone else will get make it so we can buy a couple  ;D

Best we can offer is a video of the adapter working with the reader(in case you're even interested in that), only 3-4 months of production time after the kickstarter and I can personally offer a money back guarantee if it doesn't work.