With a theoretical limit of 1.2 gigabytes a second, would a CF-to-HDMI cable adapter be an option to look into? I am working with a computer hardware engineer in Irvine on getting a functional CF-to-SATA prototype up and running, and the HDMI idea came to mind.
Pros to the CF-to-HDMI interface would be the aforementioned 10gbit/sec limit, but also the ability to record straight to a number of pre-existing devices, such as the Atomos Ninja2. Doing so, to the Ninja specifically, would give us ProRes .mov's, as well as the ability to monitor what we are recording.
Just a thought. I've already funded his research on the CF-to-SATA, and we are in the process of designing a Canon LP-E6 powered SSD dock.
Unless Atomos develop software that can encode the raw data on the fly it's not going to work, and I don't know if the Ninja even has enough processing power to do it, I doubt it.
I bought an expresscard CF reader and I'm transferring raw files from CF cards to an SSD on my macbook. The write speed to the SSD is around 150MB/s, the max speed of the CF card, the SATA interface can easily manage SSD speeds.
If we could find a way to funnel the output through the cameras HDMI-out, I find it highly plausible that we could find a manufacturer with an HDMI interfaced ARM board to team up with.
I asked Dan Keaton of Convergent Design about the possibility of using the forthcoming Oddysey7Q. His response (http://www.dvxuser.com/V6/showthread.php?308196-I-m-sorry-but-WHAT&p=1986324174&viewfull=1#post1986324174)was a non-committal "That's an interesting suggestion", but I agree with you that the manufacturers must realise that there's a big market for anyone who comes up with a recorder that will interface to the ML Raw cameras.
The CF-to-SATA/HDMI concept is a HUGELY marketable product idea. We dont need to tell them what it's for(a hack), we just need to tell them what our requirements are, and who our reachable demographic is. The market demographic includes any current DSLR with an HDMI output, and any future DSLR/Camera.
We're at the event horizon of the 4k revolution, and we're going to need faster field recorders. There are already cell phones on the international market that record 2.5k video. To keep up with the competition, camera manufacturers are going to be forced to invest in a higher output for the low-cost market. I would bet my bottom dollar that pro/am HDLSR's(t2i's-7D's) will be hitting 2.5k by the end of 2014, and introducing 4k on higher end models(6D's-CXXX's).
With this, we will see a number of recording media options: Larger affordable CF/SD's, Thunderbolt/USB3.0 tethering, or CF-Tethered recording devices.
All those hdmi clean recorders were based on Canon' s new firmware. Seriously how many would jeopdise that relationship with the manufacturer and base their profit on an openware project? Organised by amatuers with no indemnity no less. I mean really?
Quote from: 3pointedit on June 04, 2013, 12:37:12 AM
All those hdmi clean recorders were based on Canon' s new firmware. Seriously how many would jeopdise that relationship with the manufacturer and base their profit on an openware project? Organised by amatuers with no indemnity no less. I mean really?
I'm sorry if this sounds brash, but you're completely wrong. Of the various field recorders, none of them were made strictly for the Canon 5D3 1.3.1 firmware, it was the other way around. Field recorders, like the Ninja2 specifically, are strictly data collectors that compile the incoming video feed to ProRes or DNxHD from an HDMI source. This could be from a TV, BluRay player, DSLR or even an iPhone. And if you read my previous post, you'll see that I clearly stated the product could be marketed to the entire future DSLR market.
-----------ignore. double post------------------
But doesn't the effectiveness rely on a wholly proprietary (without tm) system of data and interface? A companies success would depend upon the vagaries of ML development.
Quote from: 3pointedit on June 04, 2013, 04:31:01 AM
success would depend upon the vagaries of ML development.
Hehe, you've obviously never used a Microsoft product.
Quote from: 3pointedit on June 04, 2013, 04:31:01 AM
But doesn't the effectiveness rely on a wholly proprietary (without tm) system of data and interface? A companies success would depend upon the vagaries of ML development.
This would only be true if it were an interface designed solely for ML, but it is not.
Quote from: hirethestache on June 03, 2013, 11:12:26 PM
The CF-to-SATA/HDMI concept is a HUGELY marketable product idea...We're at the event horizon of the 4k revolution, and we're going to need faster field recorders. There are already cell phones on the international market that record 2.5k video. To keep up with the competition, camera manufacturers are going to be forced to invest in a higher output for the low-cost market. I would bet my bottom dollar that pro/am HDLSR's(t2i's-7D's) will be hitting 2.5k by the end of 2014, and introducing 4k on higher end models(6D's-CXXX's).
With this, we will see a number of recording media options: Larger affordable CF/SD's, Thunderbolt/USB3.0 tethering, or CF-Tethered recording devices.
So again, there would be a large demand for some type of interface with super-speed r/w capabilities that would be compatible with current and future hardware.
Quote from: 3pointedit on June 04, 2013, 12:37:12 AM
All those hdmi clean recorders were based on Canon' s new firmware. Seriously how many would jeopdise that relationship with the manufacturer and base their profit on an openware project?
Quote from: hirethestache on June 04, 2013, 02:29:06 AM
I'm sorry if this sounds brash, but you're completely wrong...Field recorders, like the Ninja2 specifically, are strictly data collectors that compile the incoming video feed to ProRes or DNxHD from an HDMI source...
But isn't it the case that what's coming from ML is not a video stream, but a raw data stream? In order for the Atamos or Convergent Design devices to accept that data they would have to be tweaked. My guess is that it would be a pretty simple tweak, as long as the hardware is up to it, but if Canon are making disaproving noises behind the scenes, the manufactures will have to think twice before going ahead.
I think a lot depends upon whether Canon see ML Raw as a Good Thing or a Bad Thing. If the field recorder companies get tacit approval from Canon, they should easily be able to come up with something which can at the very least simply dump the ML Raw stream to disk. Looking at the filetypes supported by the Odyssey7Q, it would surely be pretty simple to add support for ML Raw. And given the resources and expertise of these companies, I doubt it would be difficult for them to develop a CF interface, if they have the will (and if nothing comes of this HDMI idea).
If the established field recorder companies don't come up with something, then it must only be a matter of time before somebody else does.
My point really was, that the manufacturer makes a product that relies upon the successful install of a non-standard 3rd party element. Perhaps they could release their own ML variant, a stable release if you will, that can guarantee performance. Otherwise how could they sell something not knowing if the camera is setup correctly? Imagine dealing with customers that don't understand the issues associated with installing ML on their camera.
And yes it would very much depend upon Canon's approval. As Magic Lantern do not earn profit from this venture (altering Canon cameras), yet another manufacturer would be. Surely there are IP conflicts here?
What if you figured out how to stage the camera to run the bin file off of the external drive or to operate the way you run a computer with SSD on it? only programs run on the SSD with the OS and you set all the data to record to other drives..... Our connection to the other drive is on the camera. Just a thought.....
Is all this talk really worth it? At the end of the day... the BMCC 4k does all that the 5D3 can do at only $400 more.
I've been testing my 5d3 raw files for a while now, and I love it. But... without an additional processor to move the data, there will be no improvements to the write speed, and therefore nothing more than 1080p raw.
Quote from: AnotherDave on June 04, 2013, 01:59:53 PM
Is all this talk really worth it? At the end of the day... the BMCC 4k does all that the 5D3 can do at only $400 more.
I've been testing my 5d3 raw files for a while now, and I love it. But... without an additional processor to move the data, there will be no improvements to the write speed, and therefore nothing more than 1080p raw.
I feel you are on the wrong board, then. Everyone here is hopeful of a stable BMCC-competitive release for our individual models.
Quote from: AnotherDave on June 04, 2013, 01:59:53 PM
Is all this talk really worth it? At the end of the day... the BMCC 4k does all that the 5D3 can do at only $400 more.
Watch it! Troll
Quote from: hirethestache on June 03, 2013, 06:53:58 AM
If we could find a way to funnel the output through the cameras HDMI-out, I find it highly plausible that we could find a manufacturer with an HDMI interfaced ARM board to team up with.
I registered just now just to suggest this idea, but you already have haha.
Anywho, as a computer engineer w/ some minor embedded systems background, I say it could be feasible if we could directly access the HDMI data stream... I haven't been able to look into the code base for ML closely yet, so I'm not sure if this is possible. I heard that the existing HDMI out is interlaced and not progressive? (unsure) But either way, we'd have to find a way to losslessly encode the RAW data into the HDMI stream, if the Digic's are fast enough... On the other end there could be hardware solution that takes hdmi to SATA or CF (might be able to try this out with my FPGA if i find it..)
Quote from: AnotherDave on June 04, 2013, 01:59:53 PM
Is all this talk really worth it? At the end of the day... the BMCC 4k does all that the 5D3 can do at only $400 more.
I've been testing my 5d3 raw files for a while now, and I love it. But... without an additional processor to move the data, there will be no improvements to the write speed, and therefore nothing more than 1080p raw.
I personally don't own a BMC, I do own a 5d3. So as it stands, the BMCC would be a few thousand more.
The BMCC also doesn't take photos, my 5D3 does.
The BMCC does NOT have a full frame sensor my 5d3 does. I'm just saying
Quote from: aaphotog on June 06, 2013, 09:10:48 PM
I personally don't own a BMC, I do own a 5d3. So as it stands, the BMCC would be a few thousand more.
The BMCC also doesn't take photos, my 5D3 does.
The BMCC does NOT have a full frame sensor my 5d3 does. I'm just saying
BMPC 4K has FF Sensor and Global Shutter...
but ur´re right. the 5d3 is a allrounder now
All this talk about BMC, let's not forget that they have yet to fulfill year old pre orders and stop theorising about how existing cameras compare to what are nothing but press releases.
Quote from: heavygrafix on June 07, 2013, 01:46:56 AM
BMPC 4K has FF Sensor and Global Shutter...
but ur´re right. the 5d3 is a allrounder now
BMCC4k has super 35 sensor, not FF
Quote from: kgv5 on June 07, 2013, 11:39:08 AM
BMCC4k has super 35 sensor, not FF
You´re right. i´ve never been so wrong
Please stay on topic: CF-to-HDMI alternative to SATA, SSD Housing & more