DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   Canon VIXIA Series AVCHD and HDV Camcorders (https://www.dvinfo.net/forum/canon-vixia-series-avchd-hdv-camcorders/)
-   -   Wearable Computer System for HD Capture (https://www.dvinfo.net/forum/canon-vixia-series-avchd-hdv-camcorders/99632-wearable-computer-system-hd-capture.html)

Kevin Kondra August 15th, 2007 11:21 AM

Here's some comparison shots. In the first one you can see the detail in the bricks and the pine trees in the background.

www.hydraclime.com/Cineform10000.png
www.hydraclime.com/HDV10000.png


The second shot show better colours in the foreground bushes.

www.hydraclime.com/Cineform20000.png
www.hydraclime.com/HDV20000.png

Andrew Plumb August 15th, 2007 11:34 AM

Kevin, assuming this is the board you're using, have you max'd out the RAM across both SODIMM slots?

Sometimes you can squeeze a bit more bandwidth out of the number crunching if you use two smaller RAM modules instead of a single larger...

Kevin Kondra August 15th, 2007 11:48 AM

Serge,

I don't understand why I would want to capture the HDMI/1080i signal with NeoHDV? Was there ever a consensus on how the HV20 processes the video signal internally to HDMI and HDV (is HDMI output just an upconverted HDV signal?)

David Newman August 15th, 2007 12:01 PM

Kevin,

If you PC can handle it, run some CineForm captures at higher quality. Medium quality was not designed for live HDMI captures, while it looks better than HDV it can look better still.

Kevin Kondra August 15th, 2007 12:27 PM

Andrew,

I'm using the AOpen motherboard without onboard RAID, similar to the one you linked. Both RAM slots are in use. The computer won't go any faster unless I overclock it, which I'll probably do next. Faster mobile CPUs are available, but I don't want to upgrade yet.

Andrew Plumb August 15th, 2007 12:49 PM

Have you tried playing around with setting CPU affinity?

The encoder may not be multi-threaded, but if you can set the capture process to one core and the encoder process to the other, you might be able to speed things up a bit more.

David Newman August 15th, 2007 01:13 PM

The CineForm encoder is N-way threaded, it use all the cores it finds for improved performance.

Andrew Plumb August 15th, 2007 01:28 PM

Thanks for the clarification, David.

...And looking back through the thread I see that Kevin did have better success with the Cineform encoder. It was just the BlackMagic codec that didn't appear to be multi-threaded. D'oh!

Serge Victorovich August 15th, 2007 02:10 PM

Quote:

Originally Posted by Kevin Kondra (Post 729422)
Serge,

I don't understand why I would want to capture the HDMI/1080i signal with NeoHDV? Was there ever a consensus on how the HV20 processes the video signal internally to HDMI and HDV (is HDMI output just an upconverted HDV signal?)

Because:
a) you have cpu usage 99% when do capture upscaled 1920x1080 10 bit stream.
b) cost of codec - NeoHDV $249 vs NeoHD $599.
c) HDMI output from HV20 is 1440x1080/4:2:2/8 bit. NO 1920x1080 10 bit ! HDV is heavy compressed mpeg2 4:2:0 8bit
NeoHDV can remove 2:3 puldown (do inverse telecine) and fully utilize output 1440x1080/4:2:2/8 bit from HV20.

But after intensive tests you already have own opinion, and money for right choice :)

Kevin Kondra August 15th, 2007 02:40 PM

Serge,

The HDMI output on my HV20 runs at 1920*1080/60i. That's part of the reason to record the HDMI output, aside from the higher quality 4:2:2/8 bit color. Only the HDV is 1440*1080.

Serge Victorovich August 15th, 2007 03:10 PM

Quote:

Originally Posted by Kevin Kondra (Post 729518)
Serge,

The HDMI output on my HV20 runs at 1920*1080/60i.

I have different information from people who have measuring devices :)
Its same mistake as 10 bit output from HV20 and 10 bit input of Intensity.

David Newman wrote on blog: "Other HV20 misinformation : when recording the Canon HV20 to tape, the image is 1440x1080, that is the HDV standard used. It is not 1920x1080, you only get that out of the HDMI port, and even then the image is likely upsized from an internal 1440x1080 image (which is still very nice.) The 1920x1080 native image is only available in the still camera mode."

Quote:

That's part of the reason to record the HDMI output, aside from the higher quality 4:2:2/8 bit color. Only the HDV is 1440*1080.
We have different preferences:) For me is color 4:2:2 with 1440x1080 better than 4:2:0 with 1920x1080. Also i prefer 1280x720p60 to 1920x1080i60.

Kevin Kondra August 15th, 2007 04:02 PM

Yes, thank-you for pointing this out for me. It's likely the HV20 upsamples a 1440*1080 image to 1920*1080 for HDMI output. I found Heinz Bihlmeir's review of this done in March. Luckily this limitation is on the camera, not on my computer.

Serge Victorovich August 15th, 2007 04:13 PM

Kevin, i've seen screenshots. You do not set "remove pulldown" option on HDlink preset for 24p?

Ian G. Thompson August 15th, 2007 05:58 PM

Quote:

Originally Posted by Serge Victorovich (Post 729533)
I have different information from people who have measuring devices :)
Its same mistake as 10 bit output from HV20 and 10 bit input of Intensity.

David Newman wrote on blog: "Other HV20 misinformation : when recording the Canon HV20 to tape, the image is 1440x1080, that is the HDV standard used. It is not 1920x1080, you only get that out of the HDMI port, and even then the image is likely upsized from an internal 1440x1080 image (which is still very nice.) The 1920x1080 native image is only available in the still camera mode."



We have different preferences:) For me is color 4:2:2 with 1440x1080 better than 4:2:0 with 1920x1080. Also i prefer 1280x720p60 to 1920x1080i60.

If you read on David only suggests that it "might" be 1440x1080 through HDMI (blown up to 1920x1080). It is still unknown whether it is true 1920x1080 from the sensors to HDMI out. Personally I believe it's the latter.

Joseph H. Moore August 17th, 2007 10:09 AM

When considering the whole 1920 vs 1440 debate, it is important to remember that the 1440 pixels are video pixels (i.e. taller than wide) and the 1920 pixels are computer pixels (square.) Remarkably, when you correct the aspect ratio 1440 becomes 1920 ... crazy how that works, huh? ;-)

My point is, you're looking at the relative loss of horizontal detail equivalent to a modest anamorphic squeeze, which is minimal.

That said, after monitoring a test image live via HDMI on a 37" 1080P monitor, and knowing that the HV20's sensor captures and even larger image, I'm left with the opinion that the HV20 is not uprez'ing on the way out to HDMI -- the image is just too clean and sharp. It's really quite amazing how much better the image is before being mangled by HDV ... enough to convince me that I had to buy and Intensity and build a capture station.

Serge Victorovich August 21st, 2007 01:22 PM

Quote:

Originally Posted by Ian G. Thompson (Post 729632)
If you read on David only suggests that it "might" be 1440x1080 through HDMI (blown up to 1920x1080). It is still unknown whether it is true 1920x1080 from the sensors to HDMI out. Personally I believe it's the latter.

DIGIC DV II processes the HD signal at 1440 x 1080 with 4:2:2 color sampling.

Solomon Chase August 22nd, 2007 10:38 PM

Kevin, could you email me? my email is sc "at" batteryfire.com. I need to get in contact with you and don't have your email address.

Wayne Morellini August 25th, 2007 04:32 AM

I thought there was an link here to an thread on an external PCIE box for laptops, but I can't find it. Here is another, and hopefully much cheaper:

http://www.theinquirer.net/?article=41876

Serge Victorovich August 25th, 2007 09:48 AM

Thank you, Wayne. Good find!

Richard Leadbetter August 25th, 2007 11:47 AM

Quote:

Originally Posted by Serge Victorovich (Post 734152)
Thank you, Wayne. Good find!

The Advanced Dock that's compatible with a whole range of Core 2-powered IBM/Lenovo notebooks has a PCIe slot...

Andrew Plumb August 25th, 2007 07:07 PM

There's the Magma ExpressBox we were discussing elsewhere, if that's what you're referring to. That'll fit up to an x16 board, but only at x1 speeds.

Ian G. Thompson August 25th, 2007 09:21 PM

Quote:

Originally Posted by Serge Victorovich (Post 732064)

Serg...it seems the link you gave me was referring to the XH A1. The following information is in the same read:
The XH A1 is equipped with three 1/3" Native 16:9 CCDs (1440 x 1080 each) -- the same image sensor that is at the heart of Canon's exceptional XL H1. They deliver outstanding picture quality, highly accurate color reproduction, and a wide dynamic range with virtually no color noise.

It says the native sensors on those cams(XH A1 and H1) are 1440x1080 each (3CCD's) while we know the HV20's sensor is a native 1920x1080 CMOS. What this means I really don't know...but it seems to me they are definately different. Also the DIGIC DV ll HD Image preocessor is used in all the cams I believe. Excuse my ignorance but how does that work acually. Is it processed to tape this way? Since the A1 does not have HDMI does it have a SDI (not sure if it's the proper term) out? And with the HDMI on the HV20's CMOS is that before the DIGIC processing...or after? It's still unclear...It doesn't matter to me anyhow...all 3 cams produce stunning pictures.

Serge Victorovich August 26th, 2007 04:05 AM

Ian, you are right about the same DIGIC DV II processor used in all Canon's HDV camcorder.
Only difference is firmware and quantity of buttons on the body of cam:)
To have idea about what is possible to do with proper hack read this:
http://mikey.wordpress.com/2007/01/0...w-format-hack/

...and google for more amazing info:)

Do you think the A1 is better (in image quality) than HV20 ?;)

http://fxsupport.de/pic/07/08/11/k02.jpg

http://fxsupport.de/25.html

Native resolution HV20's cmos image sensor not 1920x1080, but 2048x1536
With properly hack you can get 2K from 1/2,7" sensor:) Yes, its not 2/3" as altasens in SI2K but have good potention
if capture to CineformRAW straight from smos image sensor.

Wayne Morellini August 26th, 2007 08:10 AM

Quote:

Originally Posted by Andrew Plumb (Post 734391)
There's the Magma ExpressBox we were discussing elsewhere, if that's what you're referring to. That'll fit up to an x16 board, but only at x1 speeds.

Andrew, where was that, I wanted to post the link there. The Magma Box is expensive.

Serge Victorovich August 26th, 2007 08:26 AM

I hope GetCatalyst is less expensive than MagmaBox.

BTW, original "Wearable Computer System" based on mini-ITX GM965 looks as better solution than games with these adapters for laptop, imo:)

Andrew Plumb August 26th, 2007 09:32 PM

Wayne, the Intensity Capture Station thread was where the discussion came up before. (Had to look at my own posting history to find it. :-) Agreed that it's expensive, but cost is relative. You're paying for the convenience of being able to use the one board across multiple machines and/or with a laptop.

Here's to hoping more companies come out with more offerings across the board!

Wayne Morellini August 27th, 2007 07:06 AM

Serge, Yep, that adaptor card is just what I was just thinking. This excites me.

If there is an PICO/NANO itx MB with Expresscard, and angle adaptors to turn everything into an small package, it could all fit into the palm of our hands. How much processing power is needed to do basic compression?

If only they had an way to send the sensor image as an grey scale across HDMI, that would be good for compassion. I have brought this issue up before with other cameras. I suspect some cameras do have sensor test modes like this.

Richard Leadbetter August 27th, 2007 07:12 AM

Take away MJPEG and CineForm and your only choice for HD capture is Huffyuv. This still uses a fairly large amount of CPU, so the chances of using a Pico or Nano-ATX set-up are somewhat remote.

GetCatalyst also requires a 12v power supply.

Wayne Morellini August 27th, 2007 07:13 AM

Andrew, thanks for that, expensive relative to something that could do the same job cheaper. I should had realised about the adaptor Serge posted sooner. We have an bright chap in the local Office W and he was commenting about how you couldn't get an 3D card running effectively through Expresscard unless the laptop was designed to disable it's internal graphics, but I didn't find any 3D Expresscards.

Richard Leadbetter August 27th, 2007 07:28 AM

Additional graphics cards have been running in that IBM/Lenovo PCIe Advanced Dock I linked to earlier, allowing multiple monitors to be used from one laptop.

Wayne Morellini August 27th, 2007 07:30 AM

Quote:

Originally Posted by Richard Leadbetter (Post 734978)
Take away MJPEG and CineForm and your only choice for HD capture is Huffyuv. This still uses a fairly large amount of CPU, so the chances of using a Pico or Nano-ATX set-up are somewhat remote.

GetCatalyst also requires a 12v power supply.

I wonder what the actual figures are? I am starting to wonder if UMPC, or Car PC's, would have enough processing power for compression.

Andrew Plumb August 27th, 2007 08:10 AM

Quote:

Originally Posted by Richard Leadbetter (Post 734978)
Take away MJPEG and CineForm and your only choice for HD capture is Huffyuv. This still uses a fairly large amount of CPU, so the chances of using a Pico or Nano-ATX set-up are somewhat remote.

GetCatalyst also requires a 12v power supply.

That's actually a really good thing, being 12V powered. In a Pico/Nano-ITX (not ATX) system you still have to supply it with an ATX-like power supply. That means you'll have regulated 12V (and regulated 5V) at hand to power the usual suspects.

There are lots of PicoPSU and automotive PSU options that can be adapted to run off a wide range of battery voltages and capacities.

Joseph H. Moore August 27th, 2007 09:52 AM

Sheervideo is a codec that has traditionally been pretty easy on CPU's. Of course, uncompressed is really easy on the recording CPU ... but the need for a 3-5 drive array kinda negates that!

Wayne Morellini August 28th, 2007 04:39 AM

Quote:

Originally Posted by Andrew Plumb (Post 735006)
That's actually a really good thing, being 12V powered. In a Pico/Nano-ITX (not ATX) system you still have to supply it with an ATX-like power supply. That means you'll have regulated 12V (and regulated 5V) at hand to power the usual suspects.

There are lots of PicoPSU and automotive PSU options that can be adapted to run off a wide range of battery voltages and capacities.

Some of the PSU modules we found for x itx are long and thin, smaller then an memory stick in size.

Joseph, what are the actual CPU requirements for 25p/50p at 720p/1080p for sheer, mpeg and cineform etc. That sorts of gives an idea to judge things off of. At the moment UMPC's can be above 1Ghz, and 2Ghz must not be far away. When combined into an UMPC with 5-10inch display, makes for good external monitor/recording/control (Firewire/USB) solution.

Serge Victorovich August 28th, 2007 07:03 AM

Wayne, if you want small dvr for capture to Cineform 1080p24 via Intensity think about santa rosa platform, fsb800, merom T7500-T700...
I don't know any pico, nano motherboards with fsb800 to put T7700.
Real solution today based on mini-ITX like this: http://www.ibase.com.tw/mi910.htm

Socket 478 Core™2 Duo Mini-ITX Motherboard with Intel® GM965 Chipset

Cineform is the best balanced codec with less CPU usage than any other.
Future is CineformRAW ported to FPGA!;)
No need to build the hot and power hungry wintel box:)

Wayne Morellini August 28th, 2007 09:15 AM

Exactly right, that is why I am looking at lower powered platform compared to 965. Along as we have enough processing performance, it is alright, we can put something together in an day, instead of waiting for an FPGA, that might be free, or not, and then putting in an card device?

If people were serious they could start designing an small bigger than cigarette sized capture box today rather than wait for cineform FPGA solution. Using HDMI->two Analog devices JPEG2K codec chips, and usb to IDE chip, or straight to usb.

David Newman August 28th, 2007 09:20 AM

Here is the problem with that solution. Have you tried playing back HD JPEG2K files, they play at 5-6 fps. JPEG2K has never taken off as an editing format based it is too compute intensive.

Wayne Morellini August 28th, 2007 10:13 AM

OK, but I thought that was really an matter of hardware acceleration support (standard formats have everything, and JP2K is on it's own). How do you guys handle your wavelet encoding/decoding? There must be somebody working on an tricked out 2K encoder/decoder.

Now that I have got you here, any chance of an cheap capture solution like what I am talking about (ie based on free FPGA design or cheap chip) I think that BM could sell an heap at $249+ the price of your software?

Back tracking to the easier solutions, any chance that cineform 10 or 8 bit intensity 720p25/50 capture could work on an nano/pico itx board, what Ghz do you think would be required?

I might point out to people, that the 2K suggestion is only for an quick cheap dirty solution, it does not equal cineform at the same data rates.

David Newman August 28th, 2007 10:35 AM

Quote:

Originally Posted by Wayne Morellini
How do you guys handle your wavelet encoding/decoding? There must be somebody working on an tricked out 2K encoder/decoder.

There is nothing J2K developer can do to speed it up, other than waiting for faster PC. Our design has wavelet transforms are 2-3 times faster, and our entropy coding even is faster still. We see 4-6 times speed difference between CineForm and JP2K. And of course faster PC benefit CineForm too so J2K is not catching up. ;)

Quote:

Now that I have got you here, any chance of an cheap capture solution like what I am talking about (ie based on free FPGA design or cheap chip) I think that BM could sell an heap at $249+ the price of your software?
Yes, there is a chance. :)

Quote:

Back tracking to the easier solutions, any chance that cineform 10 or 8 bit intensity 720p25/50 capture could work on an nano/pico itx board, what Ghz do you think would be required?
720p25 is easy, most of the nano/pico solutions could do it. Whereas 720p50/60 is like 1080p24/25/30/i50/i60 in processing load. You will needing 2+Ghz parts.

Richard Leadbetter August 28th, 2007 10:41 AM

Wayne, you'd be lucky to capture and compress 720p using a 1GHz CPU. David obviously knows better but I think you're seriously pushing it ;)

In fact there's a whole host of obstacles to overcome - some insurmountable without a significant firmware upgrade on the board.

Firstly, Blackmagic outputs its own special brand of YPrPb 4:2:2 that barely any codec natively accepts. So you'll need to engineer your own capture tool that converts BMD's colourspace into conventional YUY2, just as CineForm has done. Then once you've done that, if you're capturing using Sheervideo or Huffyuv you'd need a RAID array for 720p/60 or 1080/24+.

There's also the small matter of Intensity's technical limitations. The only 720p modes other than 720p/60 and 720p/59.94 are slowmotion modes. So the 720p/24 option is closed off, I think.

The fact is there's a ready-made, brilliant quality solution out there and it's called CineForm HD. So I'd seriously consider Serge's Santa Rosa plan as it's the best - and certainly the most realistic - option.


All times are GMT -6. The time now is 04:40 PM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2025 The Digital Video Information Network