OHMYGAWD! Uncompressed BlackMagic 10bit recorder for $345

Yup - saw this a few minutes ago... it also looks like it covers all flavours of HD signal over HD-SDI & HDMI (which is also 1.4) so maybe the first out of the block for Sony's TC over HDMI?

The studio version (with LCD) looks very nice too.
 
Once disk storage drops in price enough for the masses, and the file sizes of uncompressed are a non issue. Uncompressed will then be the preferred file format. For some people this already is, and with products coming out like the Gemini 444 and the inexpensive Hyperdeck Shuttle people will most certainly find a way to make that workflow work. I predict external recorders will be around until uncompressed is a standard natively in camera. External recorders will adapt and change faster than the cameras, thus will always be one step ahead of most native codecs until they all reach the uncompressed holy grail.
 
Problem is, uncompressed is only a "holy grail" to a tiny segment of the market. Uncompressed is the enemy of turnaround time. I think if you asked news shooters or wedding shooters or event shooters or local commercial producers if they want uncompressed, they'd probably laugh until they passed out from asphyxiation. :thumbsup:

Even live production and EFP, uncompressed is more of an inconvenience than it would be a help. Uncompressed is for visual FX guys first and foremost, and large-budget cinema workflows that can wrangle the amount of data it takes.

Sure, storage costs will continue to come down, but when it might take you six or eight hours of realtime to run off a copy of an hour's worth of footage -- that's just not practical, for many types of workflows. For dedicated cinema shooting, fine. For big-budget TV commercials, definitely. For the local guy doing a same-day edit of an event or a wedding, or a quick-turnaround corporate image profile, no way.

I think uncompressed is pretty much exactly where it needs to be right now -- in a low-cost external recorder. I don't see the major manufacturers making any move towards uncompressed any time soon, as long as events/sports/news/broadcast/ENG make up such a big part of the market. But with the ability to add uncompressed to any system, now everyone can get exactly what they want.
 
Problem is, uncompressed is only a "holy grail" to a tiny segment of the market. Uncompressed is the enemy of turnaround time. I think if you asked news shooters or wedding shooters or event shooters or local commercial producers if they want uncompressed, they'd probably laugh until they passed out from asphyxiation. :thumbsup:

Even live production and EFP, uncompressed is more of an inconvenience than it would be a help. Uncompressed is for visual FX guys first and foremost, and large-budget cinema workflows that can wrangle the amount of data it takes.

Sure, storage costs will continue to come down, but when it might take you six or eight hours of realtime to run off a copy of an hour's worth of footage -- that's just not practical, for many types of workflows. For dedicated cinema shooting, fine. For big-budget TV commercials, definitely. For the local guy doing a same-day edit of an event or a wedding, or a quick-turnaround corporate image profile, no way.

I think uncompressed is pretty much exactly where it needs to be right now -- in a low-cost external recorder. I don't see the major manufacturers making any move towards uncompressed any time soon, as long as events/sports/news/broadcast/ENG make up such a big part of the market. But with the ability to add uncompressed to any system, now everyone can get exactly what they want.

I do agree with what you are saying. There most definitely is still a need to compress for most. You have to think tho, someday IO speeds and CPU speeds will also be a non issue. It's still the news guys holy grail, it's just a bit farther out.
 
Oh

my

god.

I just saw this for the first time. Essentially, I can bypass the h264 compression on my DSLR with this bad boy?????
 
The manufacturers have 'crippled' the signal coming out of the cameras intentionally, otherwise there would be less of a need to buy the more expensive cameras made by the same companies. The HDMI out is useful for external monitoring.

External recorders dont work; otherwise everyone would have been using DSLRs with Nanoflashes for the last number of years.

Fewer people would buy an Af100 if they could get a GH2 with an external recorder instead.

The day someone releases a DSLR with a clean HDMI out is the day I jump on the bandwagon. In the meantime you will have to cough up for the Af101, the F3 or the FS100 instead, which are completely different price brackets.
 
Last edited:
HDMI out of the DSLR's is subject to the quality limitations of the live view processing built into the camera. Scaling resolution encoding etc. HDMI supports a lot of different digital resolutions and compression formats.
 

Just to give you an idea of how far we are from your goal, to the best of my knowledge the Sony F3 at ~ $16K US w/o lens is currently the least-expensive, >2/3" sensor, interchangeable-lens camera with live HD outputs that can make _full_ use of the recently announced/shipping crop of portable HD recorders.

Of course, this situation will change for the better when some less-expensive cam of this type gets announced, but in the meantime don't hold your breath if you've got work to do.
 
You've encouraged me even more with that statistic! Certainly I can see the difference between the internal H264 compression and the Hyperdeck thru a crippled HDMI port. So what if you can see an even greater improvement for $16k? Noone else will notice it with a PC as a monitor. Am I correct?

If your final delivery is going to be for youtube or something only seen on a small computer screen then you definitely wont be needing uncompressed footage (and the codec in your dslr will more than suffice - indeed DSLR footage has made it on to major productions like The Black Swan with the onboard codec).

You just have to be aware of the limitations (moire, aliasing, artifacting under specific circumstances) and work around them.
 
Can somebody explain what you guys mean by the 19% up-scaling when using this device with a DSLR? Thanks. Also, will there be any significant improvement when using this device with a DSLR even though there isn't a legitimate uncompressed signal?
 
Someone else said the signal isn't actually uncompressed, but even if that's true, as long as there's a significant increase in quality with the device, I'll purchase it.

So, the device captures exactly what you see in live view. Live view has black bars at the top and bottom in order to maintain a 16:9 image. You zoom-in to get rid of the black bars.

Since a 16:9 video image is letter-boxed within a more squarish screen on the 60D, does that mean the output image isn't 16:9? Rather, it outputs a squarish resolution with a 16:9 image letter-boxed inside?

I'm probably articulating this terribly, so sorry if it's confusing.
 
Last edited:
HDMI is just a pipe. It supports many resolutions and standards both compressed and uncompressed. We connect videoconference codecs to displays via HDMI and that is highly compressed video, even in HD. Uncompressed 4:2:2 1080P is 1.5 Gb per second. That is not what you are getting from any DSLR.
As long as the video meets a supported format for this recorder, it is recorded without further modification, but that says nothing about the quality of the signal you feed it.
 
Last edited:
I do not believe it is possible to compress a 4.2.2 signal. Or it wouldn't be 4.2.2 at all. Other than removing the 2 bits per pixel compared to 4.4.4. Isn't that what is coming out of the DSLR? It does not first pass thru the H.264 codec inside the camera. http://en.wikipedia.org/wiki/HDMI


Yes you can have compressed 4:2:2 ; examples of lossy 4:2:2 compression are prores, dnxhd, cineform, some mjpeg variants, avc-intra 100, other h.264 profiles,.etc...

You can have lossless 4:2:2 compression as well eg. huffyuv, ut video codec, lagarith etc...

"4:2:2" usually refers to the chroma subsampling in Y'CbCr (a color model), not the bit depth. "2 bits per pixel compared to 4:4:4" have nothing to do with each other. 4:4:4 means no subsampling in chroma, it usually refers to RGB, but you can have Y'CbCr 4:4:4 as well

10bit vs. 8bit is the bit depth per channel ; how many values in the range of expression. For example , luminance Y' in 8-bits would have 0-255 in full range. 10-bit Y' would have 0-1023 in full range

If you have a 1080i signal , that reduces the vertical resolution by 1/2 (each field is is 1920x540) , but the effective resolution may be slightly higher depending on many factors, and whether there is motion




When you zoom by 50% you can subjectively tell the difference, going from 720p to 1080p. 19% not so much. Do the experiment yourself, don't trust me. Do it with a still image JPG.
It is much less visible with a moving video, I've tried many. It is the apparent sharpness that we notice. Don't ask me to define that, I've tried previously.

19% is a substantial loss, even on lower res / softer camera. There are also framing differences when you crop & zoom
 
Actually you're right! It's 4:2:2 1080i60. Which is about 1.5Gb/s or 166GB/s or 10GB/min.

Hmmm. So that's 600GB per hour. Everyone: Uncompressed is expensive! For less than the cost of one hour of storage of uncompressed on fast SSD, you can by a Ninja! And then you can use cheap storage. Even if you use expensive SSDs, they'll still last between six and ten times longer! ProRes is optimised for editing. And it's easier to edit with than uncompressed. And it's visually lossless.

Dave Shapton
President
Atomos EMEA
 
Yes, with 4:2:2 the COLOR space is compressed, as I said, by 2 bits for each color channel.

No. Bits per channel component is a different concept than chroma subsampling.

In Signal Theory, there is more useful data in 60i than 30p. There may be a "substantial loss", but you can't see it! Try it with a JPG! Don't forget to exclude aliasing when you design the simple experiment.

60i has more temporal resolution (59.94 moments in time per second are represented) , but less spatial resolution.

There is debate on the "actual" percieved spatial resolution if you take into account Kell factor

Even if you discount the DSLR aliasing, there can be even worse aliasing introduced when viewed on progressive displays (LCD, plasma) - it depends on the deinterlacing algorithm used.




What do you mean by framing differences?

If you crop and resize back to 1920x1080, there is less peripheral information because you are "zoomed in."

eg. If you were planning a wide shot for a certain scene in a movie, you would have to factor that in before shooting




Regardless, I think this is great that there are so many choices for recorders this year
 
Can we just record at say, XDCAM level with this and just have endless recording time? I only want to maybe record at 35 or 50mbps.
 
You can compress and decompress before encoding to 4:2:2. But you cannot just compress then encode.
If you are using the hyperdeck, it's records uncompressed video. It just records whatever signal is fed to it. When people say "compress" that implies encoding . Uncompressed video doesn't need to be decompressed, so yes you can just encode to your desired format directly

But if you are using a compression format (either lossy or lossless), it needs to be decompressed (decoded) first before being fed into the encoder

The content near the edges are cropped off? I don't think so.
Go back and re-read what I said. I said "if you crop and resize back to 1920x1080". If you've already cropped it, of course the edges are cropped :)


Cheers
 
ProRes is optimised for editing. And it's easier to edit with than uncompressed. And it's visually lossless.

Dave Shapton
President
Atomos EMEA

Hi Dave,

Uncompressed video is actually a direct digital copy of your input and requires far less CPU resources than ProRes. Try playing back a stream of Uncompressed 10-bit 4:2:2 in Final Cut Pro and compare how much CPU resources you'll be using compared to using ProRes.

Kristian Lam
Blackmagic Design
 
Back
Top