I agree...................... a small laptop + PCI chassis could probably be configured as a reasonably portable capture system. I'm going to look into this idea.

IMHO, I doubt a dedicated mid-sized 'black-box' (without a display) would sell too well. But that's just MY opinion.............. I've been wrong plenty of times before. ;-)

JohnG

Id be interested in this for sure. I was just pondering what it would take to build one myself. So the issue seems to be that capturing to something like cineform HD is just past the limits of current mobile processors?
based on the requirements listed cineform's website, I can see why 60i might be a problem to capture on such a system. 2.33GHZ Core2Duo Mobile might just barely be able to compress that in real time. I wonder if David Newman from cineform has some insight about how cineform would encode on the fastest modern mobile processor. How about using a desktop CPU?

Does it take more or less CPU to perform an inverse pulldown while encoding? i imagine the operation itself takes cpu time but then thats 20% less data that has to be compressed. I suppose the CPU requirements would be for a codec like bitjazz would be less... i suppose that would put it in RAID territory. Cineform does have a pretty good sweet spot in terms of not needing too much storage speed or CPU time. Perhaps it could be tweaked to work with slightly slower CPUs?

Let me think about how much this would be worth...

Mini-ITX motherboard - $300
Laptop CPU - $650
2GB RAM - $200
80GB 2.5" Hard Drive - $80
Intensity HDMI card - $250
PSU - $200
Battery - $100
Embedded Storage - $100
Custom Chassis, Cooling - ?

So we are talking almost $2000 in off the shelf parts. Not too much Software development costs if it records using some already available software (cineform).

I guess my answer based on these numbers would be ~$3000. Although I think costs could be reduced by using a desktop CPU or increasing the formfactor slightly. Then again, those costs might resurface due to the larger battery and more complex custom cooling system.

are there companies that make expresscard to pcie external adapters? I know i remember seeing one that was expresscard to pci, which wouldnt work too well.

We could do some testing with 2.33GHz Core2Duo Mobile to give a more accurate estimate, but it might work if you limit other processing. Silicon Imaging uses the same processor for encoding 2K CineForm RAW. 1920x1080 4:2:2 8/10bit is a little more data but siginificantly less preprocessing needed. The inverse telecine (or pulldown removed) does use less CPU than the 20% saving achieved through encoding 24p vs 60i. So yes you may find you can encode 1080p24 with a mobile CPU, and no issue for today's dual core desktop CPUs. As for tweaking CineForm further, it is already optimized to the max for Intel procs, likely more than any other codec. You can control the CPU requirement by dropping the bit-rate, but that starts to impact quality if you go too far. I think it would be better to find a CPU that meets your purpose rather than limit the quality from our current levels (http://cineform.blogspot.com/2007/03...en-screen.html)

Update
 

@Mathieu
I did encourage you to email me and sign up to the private mailing list; you haven't missed anything yet but an announcement is coming in the next couple of weeks. Sorry to be so hush-hush about this but it is a competitive landscape.

I will let out a teaser though - we're working not on one but two capture solutions, HD1 and HD2; you can upgrade from HD1 to HD2 by shipping it back for an upgraded unit.
HD1 can do HDMI 4:2:2 8bit capture in a 17x17x5 cm form factor (battery included!).
HD2 can...sign up to the mailing list! :)

Any Update
 

Hey Alex

Its been some time since your last post on this subject.

Do you have any news yet?

Hi Jonty, thank you for posting. The news is...we're still working on this. I've made a decision early on not to hype the product (too much) until it is announced (vaporware and all).

I can make a few statements that are harmless enough:
- there is a team working on this
- there is no target date for series production, certainly not for this year; we do have something working, but in dev board form factor.
- we know how to make it quite small, small enough to mount the DDR underneath or on the side of the camera (external dimensions 160 x 120 x 52 mm); battery is internal, storage is external.
- we've largely finalized the BOM (Bill of Materials) and have contacted suppliers for volume pricing
- we won't be using Intensity cards, but our own custom HDMI aquisition design (this allows us to shrink the form factor, but lengthens the R&D time)
- a larger form factor using Intensity Pro can also be built around the same design (storage is internal); external dimensions are 220 x 160 x 52 mm
- the codec is Cineform Neo HD
- recording time on one battery charge is estimated at 2 1/2-3 hrs.
- it will record 1920x1080p24 4:2:2 8bit; it won't do i60, but it will do i50.
- it may be able to do pulldown removal from i60 to p24 at 1920x1080; yet to be determined.
- the enclosure is black and stylish
- it's going to run hot; the enclosure is the heatsink

Damn, this sounds good. I might very well buy one if the price is reasonable.

Pulldown removal
 

question for those in the know; we need to test 60i->24p pulldown removal for the DDR project; my acquisition options are very limited (PAL land). As it stands, we either:

1) rent a F-950 and hire a DP to operate it; go off and find some complex outdoor scenery to see if the codec holds up. More than once, difficult logistics, on rental's place schedule; cost is 500 EUR/day with the DP.

2) buy a HV-20 from an US outlet (only E models over here of course). We get to use the cam whenever we need it, and the cost is a couple of days of F-950 rental.

Aside from the 8 bit vs 10 bit resolution in F-950 (which doesn't matter as *this* version of the DDR is 8 bit), is there a compelling reason not to go with 2) ?

We're looking for scenes with lots of detail and color, as I said stress-testing the real time compression. Would one cam's 8 bit 4:2:2 HDMI signal be (significantly) different than another's? (I do expect a 10 bit signal to be different, but that's not what we'll be testing).

Looking forward to your replies; it will help us make major decisions in this project.
As a side note, how um..unfortunate would be if pulldown removal didn't work 100% of the time (like very complex scenes)? I know the situation is close with our current part choices.

Any update?

Any guess on delivery? Weight? Price?

Mathieu, I do value your continuing interest; put yourself in my place for one moment please. Yes I do have projections for those questions, they're part of the business plan. However making this information public - for a yet to be announced product - will paint ourselves in a corner. There are also features I'm not disclosing before announcement to maintain competitive advantage.

As I mentioned before, there will be no series production this year. The thing is this project has snowballed in terms of engineering to the point where it bears little resemblance to how this thread started. We gain a lot in form-factor by doing this, at the expense of lenghtened R&D.

It's a HDMI DDR and it uses Cineform, but beyond that, it is (literally) going to be a small black box, attachable to a cam like Sony V1.