Hi,

I just did some shooting with the Macbook Pro (2.16Ghz, 2GB RAM), and it's not fast enough . . . the RAM buffering helps to get the shot, but the most you can record before overflowing the RAM buffer is around 20-30 seconds.

The current Macbook Pro's are not a good solution for this camera.

Merom-based systems coming out in August should run great though.

Thanks,

Jason

Hi Jason,

What application are you using for capture if I may ask?

-gb-

We're using our own application. You can see a screen-shot on our website at www.siliconimaging.com/DigitalCinema/design.html

Now of course you can use the Macbook, I'm just noting that you're not going to get extended real-time recording capabilities like you will with a Merom-powered system.

The current Macbooks are right on the edge . . . another 5-10% CPU performance and we can do real-time recording in CineForm RAW as long as you'd like, but right now, the "umph" just isn't there, and you end up running behind, filling up the RAM Buffer.

So if people are going to invest in a notebook for capture, our recommendation would be to go with a Merom-powered system for the utmost performance and stability. Current Yonah systems will do, just have lots of RAM because you will need it for longer takes (1-2 minutes).

The only minor addendum to Jason's comments is that for RT CineForm RAW acquisition at HD Resolution from the SI camera, we're doing long-term capture using T2600 (2.16GHz) Yonah-based core-duo captures on Windows without a hitch as long as the Intel chipset is used for the GigE NIC. There's no doubt that Merom-based machines will solve all problems, but it's also true that the Sony AR190 laptops with the T2600 work very reliably.

Yeah, that's the major shortcoming of the Macbook Pro's . . . they use Marvell gig-e instead of the Intel NIC, so that added performance jump from the Intel NIC (i.e., in the Sony dock) isn't available.

the merom is the way to go. I didn't buy the yohan because it gets too hot and is just on the edge of pushing the power to the limit.

The acquiring software works also into Mac book pro? Once you've acquired the footage, you can also playback it but with what player? Quicktime?
There's some advantage acquiring with new Merom-based systems? Considering that you cannot edit on mac book pro, and you must transfer the footage to a workstation, it's better that also the workstation is Apple (new powermac intel for example) ? But you cannot edit natively cineform codec, so you must reconvert to quicktime format losing time to do the conversion, right?
So it's more reasonable to have a Pc notebook (for example Vaio or others available when camera ships) because you're sticked to Adobe Premiere that works only on Pc.
I really wanted to buy those new Macs! Sigh! I'm condemned for all my life to use pc platform...

Hideaki, CineForm RAW files will be editable on the Mac in the future, but we have not yet made this release. In addition, we will add an option in the SI capture application to acquire CineForm RAW with a QT wrapper instead of an AVI wrapper.

So...that means you'll be able to use a Mac running Bootcamp (Windows) to acquire from the SI camera. Then you can boot up into MacOS and edit the QT-wrapped CineForm RAW files. You won't have to transcode. So you'll be able to buy that Mac that you covet as long as you run Bootcamp for the D2D recording.

Merom will (likely) become the recommended minimum processor in a laptop for D2D recording. The new "Conroe" or even the older Pentium D 955-Extreme Edition are fully adequate on the desktop side for D2D recording. Apple hasn't yet announced their desktop/workstation lineup, but I suspect they'll use the new breed of Intel processors and will have no problem either recording or editing CineForm RAW.

What differencies about quality will be between a QT-wrapped CineForm RAW files (used in Final Cut Pro) and a cineform (avi?) for windows applications (adobe premiere)?
If I'll acquire with 20000 dollars camera package, I've the hard disk, so I'll have the option to write on the SEAGATE HDD QT-wrapped Cineform RAW files (they still will be at 10 bit? as the version for Premiere?)
They will have different attributes, for example variable framerates will work and possible to edit also in Final Cut Pro?
What disadvantage we'll have if we buy a mac workstation to edit cineform codec instead of a pc workstastion?
Of course I'm talkin' about new powermac intel supposed to be available in september, they will be great to edit cineform files?
I noticed that a Mac workstation cost less money (with more ram) than a pc workstation, so for me it's more convenient invest in Mac workstation.
If the new ones will work great as pc platforms, it's more convenient to buy a Mac, if We got NO Disadvantages.
I'll buy a Mac ONLY to work with Silicon Camera, so hope to have soon in your homepage a confirmation that shows that edit Silicon Imaging is possible with Mac and Pc workstations.
Sorry for my ignorance about editing software and cameras, but I'm waiting to buy my first one and Silicon looks promising to be my first camera.

Importantly, there will be zero visual difference between CineForm Intermediate files, whether they're wrapped in AVI or QT. So that will not be an issue.

We haven't made any final determination at this point about supported features within Final Cut. What we do know is that our "hooks" into FCP are very different than they are into PremPro, so the editing experience will be different because of this. Further, I suspect we'll roll out features into FCP over time. We will not have a feature set as comprehensive as PremPro when we initially launch FCP support. But our first support WILL include 10-bit because we want to support the SI camera on FCP as soon as reasonably possible.

I'll also volunteer that we have QT-wrapped CineForm Intermediate files, created on a PC and copied to a Mac, now working in the QT player under MacOS. Importantly, we've proven interoperability between OSs. But we're still working on integration into Mac apps like FCP, AE, Motion, etc.

"I just did some shooting with the Macbook Pro (2.16Ghz, 2GB RAM), and it's not fast enough"

Haha. Really? The MBP is a fast machine. What is the software written in, Javascript?

Ben,

The background on Jason's comment is that we're performing real-time CineForm RAW compression in software on HD-resolution raw sensor data using a mobile processor in software. The fact that we can do this at all on a laptop, without specialized hardware, is really unprecedented. The resulting workflow offers a combination of higher visual fidelity (more dynamic range - all the stuff this thread has covered) and a real-time online editing environment.

Besides the visual quality emphasis, the code for CineForm RAW compression has man-years of effort invested in low-level SSE2 coding and n-way threading. Before Merom this level of performance required either a desktop machine or specialized hardware. Prior to the last year such performance required a workstation-class machine.

Kudos to Intel for delivering the CPU power and memory bandwidth to allow us software guys to extract such performance.

David.

The MBP is a fast machine. What is the software written in, Javascript?

No Ben, it's not written in Javascript . . . there's quite a bit more happening in the software than just compressing to CineForm, which in and of itself is quite a feat considering that you're doing real-time wavelet compression on a laptop that only last year required a dual-processor Opteron workstation (252's or higher).

BTW, I've started doing testing with Merom, and it's definitely fast enough (the T7400 or T7600).

Also consider the fact that I'm doing testing of the HIGHEST quality level of compression from Cineform (so the most CPU usage) on our most stressful settings in the camera . . . that means higher gains, highest resolution the camera can deliver, etc. I don't think it would be fair to "qualify" a laptop but basically use the lowest-quality compression settings, low-resolution (like 720P), etc. I'm putting it through it's paces, taking it through the ringer, and seeing what comes out the other side. If it doesn't make it, it doesn't make it. But as mentioned, the good news is that Merom processors are great, running fast, and will be on the market in less than a month.

We'll have both "minimum configurations" and "recommended configurations".

Minimum means just that-minimum . . . you're only going to get so far before you exceed the requirements of the laptop.

Right now the "bare minimum" requirements for 1080/24P are a 2.0Ghz Yonah and 512MB of RAM in a dual-channel configuration (that's very important). A single-channel configuration of 512MB RAM (meaning only one 512MB stick), has not been sufficient enough to-do the trick at all. So, if you do have a current Macbook Pro, it *will* work . . . just not at the maximum-quality settings.

Our "recommended" configuration is a top-of-the-line Merom system (T7600) with 2GB+ of RAM . . . the more, the merrier.

Hey Jason please read my new post:
http://www.dvinfo.net/conf/showthread.php?t=73279

Also consider the fact that I'm doing testing of the HIGHEST quality level of compression from Cineform (so the most CPU usage) on our most stressful settings in the camera . . . that means higher gains, highest resolution the camera can deliver, etc. I don't think it would be fair to "qualify" a laptop but basically use the lowest-quality compression settings, low-resolution (like 720P), etc. I'm putting it through it's paces, taking it through the ringer, and seeing what comes out the other side. If it doesn't make it, it doesn't make it. But as mentioned, the good news is that Merom processors are great, running fast, and will be on the market in less than a month.

We'll have both "minimum configurations" and "recommended configurations".

Minimum means just that-minimum . . . you're only going to get so far before you exceed the requirements of the laptop.

Right now the "bare minimum" requirements for 1080/24P are a 2.0Ghz Yonah and 512MB of RAM in a dual-channel configuration (that's very important). A single-channel configuration of 512MB RAM (meaning only one 512MB stick), has not been sufficient enough to-do the trick at all. So, if you do have a current Macbook Pro, it *will* work . . . just not at the maximum-quality settings.

Our "recommended" configuration is a top-of-the-line Merom system (T7600) with 2GB+ of RAM . . . the more, the merrier.

Howdy Jason...
Jumping over here after a lengthy chunk of time on the HVX200 forum (my current "wait til things shake out" camera). Pardon the fact that some of these questions are probably answered elsewhere... I've been through a lot of material, but sometimes it hasn't been entirely clear.

I'm gearing up to embark on my first theatrical-release venture, a feature-length documentary. A fair amount of the B-roll will probably happen on the HVX (for portability and onboard audio), but I'd prefer to shoot all the interviews on something with greater image quality and DOF control (the curse of 1/3" chips). There will also be footage shot of theatre performances, where greater image quality and latitude would be helpful. The SI is definitely a top contender, despite my concerns about the timetable of FCP/Avid compatibility.

I'm platform-agnostic in terms of my day-to-day work for other clients, but would prefer to stick with Mac hardware for my own equipment. Since I've raised the funds to produce the film (but not to spend a large chunk of it on equipment purchases), I'd likely be looking at just the camera head. It sounds like both the head and the next generation of MacBooks may be available soon enough to consider using this system for this project. Since I plan on having an audio tech for all interviews/theatrical work anyway, recording two-system audio to a field recorder isn't a major stumbling block.

Soooo... a few questions:
When you say the Merom MacBooks will be sufficient to operate as the recording device, are you talking about 1080/24, or all available frame rates up to 72fps?

Since you're describing the port over to Quicktime as being a rewrap, I'm assuming the Cineform is a container-type file structure? If so, is it an atom-type file to maintain as much file integrity as possible in the case of data failure? (Disclosure: I know only ~just enough~ about codecs and file structures to be dangerous...but I was married to a mainframe programmer/Java architect and understand the iterative-commit process of the atomic file structure. I'm also familiar with the basics of wavelet technology, from a prior life as a still photographer.)

Since the heavy lifting in the laptop is in processing the raw signal, I'm assuming there's no significant loss of overhead in writing the files to an outboard FW400 drive rather than the laptop's internal drive? Or does the FW drive need to hang off the faster Express/34 cardslot?

I'm seeing 12 MB/s discussed as the data rate. Assuming that's not a typo (MB instead of Mb), that's 96 Mb/s, which would outstrip even a 7200rpm drive. So is the data rate really only 12 megabits/sec? Or am I mixing up the final deBayered file size with the wavelet file size?

Are there labs out there already geared up to do film-out from the raw Cineform files, or would I need to output to D-5 (or transcode the footage to DVCProHD or 10bit uncompressed QT?) This gets even trickier, since I'll be mixing Cineform with the HVX DVCProHD. It seems like transcoding it all to DVCProHD would make sense, but do the transcoded files look as good as the originals?

I'm seeing the camera body drives described as hot-swappable. Does it hold a pair of 2.5"s, or is there simply enough onboard RAM buffer to allow a quick swap of drive cartridges (of course, it's debatable whether you could even do that without shaking the camera to an unacceptable degree...)

Are the drives accessible without actually opening up a panel on the camera body? From the photos, it doesn't appear so.

If I remembering correctly, the announced pricing of the camera head includes "two C-mount Fujinons." Is that optional for those of us who'd rather look at getting a single, high-end S-16 lens? And does the powered camera head (without camera body & brick) provide power to an HD video lens's zoom servo?

Finally.. (yeah... really.. I know.. lots of questions) one thing that drives me nuts on the HVX200 is the inability to leave the focus assist turned on at the viewfinder while shootoing. When I'm working with a monitor, I can use the monitor for framing, but would like being able to judge critical focus through the entire shot if I have a fidgety interview subject. Is that something you guys can accomplish (BTW, 2x really isn't enough to judge critical HD focus...)

Soooo... a few questions:
When you say the Merom MacBooks will be sufficient to operate as the recording device, are you talking about 1080/24, or all available frame rates up to 72fps?

It should work for all available frame-rates.

Since you're describing the port over to Quicktime as being a rewrap, I'm assuming the Cineform is a container-type file structure?

Yes, but there's no "re-wrapping" . . . you will record using the DirectShow interface directly to a QT file format. Of course if the file gets "damaged", and not closed properly, then it's not readable, although we have some workarounds for not loosing all your data through segmenting the files during a long record

Since the heavy lifting in the laptop is in processing the raw signal, I'm assuming there's no significant loss of overhead in writing the files to an outboard FW400 drive rather than the laptop's internal drive? Or does the FW drive need to hang off the faster Express/34 cardslot?

Outboard firewire should be perfectly fine. We've been using USB 2.0, and it's been working great.

I'm seeing 12 MB/s discussed as the data rate. Assuming that's not a typo (MB instead of Mb), that's 96 Mb/s, which would outstrip even a 7200rpm drive. So is the data rate really only 12 megabits/sec? Or am I mixing up the final deBayered file size with the wavelet file size?

Actually laptop hard-drives are easily able to sustain that data-rate. We've been using 5400RPM laptop drives which go from 35MB/s at the start to 18MB/s at the end, and it's all working fine. And yes, the datarate is 12MB/s, or 96Mb/s.

Are there labs out there already geared up to do film-out from the raw Cineform files, or would I need to output to D-5 (or transcode the footage to DVCProHD or 10bit uncompressed QT?) This gets even trickier, since I'll be mixing Cineform with the HVX DVCProHD. It seems like transcoding it all to DVCProHD would make sense, but do the transcoded files look as good as the originals?

Yes, Pixel Harvest can do film-outs from native Cineform files. Don't transcode to DVCProHD, that will ruin the resolution (since it's not a full-raster codec) and add compression artifacts.

I'm seeing the camera body drives described as hot-swappable. Does it hold a pair of 2.5"s, or is there simply enough onboard RAM buffer to allow a quick swap of drive cartridges (of course, it's debatable whether you could even do that without shaking the camera to an unacceptable degree...)

It's hot-swappable, but not in that sense. You can store up to 4 hours or more on a 200GB hard-drive. That should be plenty for an interview.

Are the drives accessible without actually opening up a panel on the camera body? From the photos, it doesn't appear so.

Yes, the drives slide out the back (in that photo you see).

If I remembering correctly, the announced pricing of the camera head includes "two C-mount Fujinons." Is that optional for those of us who'd rather look at getting a single, high-end S-16 lens? And does the powered camera head (without camera body & brick) provide power to an HD video lens's zoom servo?

We don't have pricing for PL-mount lenses, you will have to supply those yourself. Also we don't support B4-mount lenses since we've found they aren't apochromatic, so that can cause problems with a single-sensor design.

Finally.. (yeah... really.. I know.. lots of questions) one thing that drives me nuts on the HVX200 is the inability to leave the focus assist turned on at the viewfinder while shootoing. When I'm working with a monitor, I can use the monitor for framing, but would like being able to judge critical focus through the entire shot if I have a fidgety interview subject. Is that something you guys can accomplish (BTW, 2x really isn't enough to judge critical HD focus...)

Yes, we have a "loupe" mode that you can leave up during recording, and it does a really, really tight zoom (like a loupe). With it you will know if you're in critical focus or not.

Hope this helps,

Jason

snip

Actually laptop hard-drives are easily able to sustain that data-rate. We've been using 5400RPM laptop drives which go from 35MB/s at the start to 18MB/s at the end, and it's all working fine. And yes, the datarate is 12MB/s, or 96Mb/s.

snip

Hope this helps,

Jason
Indeed it does; thanks a lot!
As it turns out, the material I'd been reading about typical ATA data transfer rates were using Mb to mean megabyte instead of megabit. Thus the confusion. It was a whole lot easier when all I had to worry about was ASA...

You're really finding that good HD glass mas too much chromatic aberration? I'll admit I'm surprised.

You're really finding that good HD glass mas too much chromatic aberration? I'll admit I'm surprised.

It's not the problem that it's not "good glass", it's the problem that the lenses were not designed to properly focus all three color planes on the same physical plane . . . they are specifically constructed to slightly offset the RGB channels so that they compensate for the different refractive indexies of the prism as the light wavelengths are split into the three RGB primaries. As a result, putting a lens designed in such a manner in front of a single plane sensor where the RGB wavelengths need to be focused at the same physical plane creates slight offsets in the RGB channels that will contribute to chromatic abberations. So you need a lens, like a film lens, that was designed to focus the individual RGB channels all at the same physical plane.

It's not the problem that it's not "good glass", it's the problem that the lenses were not designed to properly focus all three color planes on the same physical plane . . . they are specifically constructed to slightly offset the RGB channels so that they compensate for the different refractive indexies of the prism as the light wavelengths are split into the three RGB primaries.

I had never run across that interesting little piece of information. Makes sense optically... though, given that each manufacturer's prism assembly is different, you'd think that'd be corrected in the camera head, not by expecting the lensmaker to design corrective optics (that are expected to work on any HD cam, not a specific model.)

Nevertheless, I understand what you're saying. Interesting.