220 Mbps I-Frame-Only In Next Release

Dan Keaton · June 9th, 2009, 01:46 AM

Dear John,

The exact performance of the 220 Mbps I-Frame Only is still to be determined.

In other words, we do not know which will be better in terms of absolute image quality, ignoring the size of the files.

We feel that they will be close, but it is too early to tell.

The "Holy Grail" is a codec that produces an image that is visually indistinguishable from uncompressed, and produces the smallest file sizes.

Our 100 Mbps Long-GOP mode, in our tests, and in tests performed by others, meets this standard.

But, some want an I-Frame Only (Intra-Frame) option, so we offer both.

The 100 Mbps option is what I recommend, as this has been proven in many tests, and it produces very compact file sizes (for high quality HD images). The 220 Mbps I-Frame Only file sizes are larger, thus recording time is reduced.

Bill Ravens · June 9th, 2009, 06:28 AM

Hi Dan...

I think we've been down this road before. But, at the risk of being redundant, I'd much rather see 10-bit encoding than 220mbps I frame.

Piotr Wozniacki · June 9th, 2009, 06:39 AM

Quote:

Originally Posted by Bill Ravens

Hi Dan...

I think we've been down this road before. But, at the risk of being redundant, I'd much rather see 10-bit encoding than 220mbps I frame.

Dan and Mike,

I'd like to strongly express my opinion Bill is 100% right... If only it's in the realm of the NanoFlash basic design, the addition of (even chargeable) upgrade to the long GOP (or uncompressed, as MPEG2 theoretically doesn't provide for this) 10 bit 4:2:2 is MUCH more appealing that even the highest Mbps I-frame encoding!

Dan Keaton · June 9th, 2009, 09:29 AM

Dear Bill and Piotr,

We fully understand, we would love to do what you want.

All MPEG-2 is 8-Bit. All HD Television is 8-Bit.

Even if we made a custom version of MPEG-2, the NLE systems would have to support it for it to be really useful.

Our Flash XDR will, in the future, with an extra cost upgrade support 10-Bit full uncompressed. This is possible since the Flash XDR has four CompactFlash card slots.

We calculate, that using today's CompactFlash card technology, it takes writing to four CompactFlash cards simultaneously to support full uncompressed. The nanoFlash has two CompactFlash card slots.

Bill and Piotr, I understand that you know the next point, but some don't:

For us to effectively record 10-Bit, a 10-Bit camera needs to be used. There are some 10-Bit cameras at the ultra high-end, and some low cost cameras have 10-bit, but most are only 8-Bit.

Just as soon as we are able to provide a 10-Bit solution, we will do so.

Richard Welnowski has performed the following test:

1. Record using a great 10-Bit camera, the Thomson Viper, in full 10-Bit 4:4:4.

2, Record simultaneously to our Flash XDR at 100 Mbps Long GOP.

3. Compare each image, frame by frame. The best way I know of to do this is a "Difference Mask" in Photoshop.

The results were that the differences were very minor.

Some are concerned about banding. We have not seen an instance of banding when using our 100 Mbps Long GOP mode. Most attribute banding to 8-Bit, but it can also be created when using a lower performance codec.

No one who has ever tested the Flash XDR or the nanoFlash has ever complained about the image quality or reported banding.

I hope this helps.

Mike Schell · June 9th, 2009, 09:46 AM

Quote:

Originally Posted by Piotr Wozniacki

Dan and Mike,

I'd like to strongly express my opinion Bill is 100% right... If only it's in the realm of the NanoFlash basic design, the addition of (even chargeable) upgrade to the long GOP (or uncompressed, as MPEG2 theoretically doesn't provide for this) 10 bit 4:2:2 is MUCH more appealing that even the highest Mbps I-frame encoding!

Hi Piotr-
Understand, 10-bit recording is possible if you don's mind a much larger size unit (not camcorder mountable) and much greater power (which means a much larger battery). In time there may be hardware options that will allow to to create a small low-power 10-bit recorder, but it's not possible with the technolgy we have access to today.

Also recall that 10-bit recording buys you nothing unless your camera outputs 10-bit. Yes, the Sony EX1/3 gives full 10-bit as well as the new Panasonic HPX-300, but nothing else in the sub $10K range is 10-bit. All HDV, AVCHD and (I suspect) DVCProHD based cameras are only 8-bit. Even many of the higher cost cameras are 8-bit as well as every HDMI camera we have tested.

10-bit recording does help, but according to a our high-end Viper users (who are shooting several movies with the Flash XDR), you simply have to know how to shoot the scene to allow for 8-bit recording.

I would agree that 10-bit recording is advantageous for output to film, but I doubt you'll gain very much (if anything) if your final output is TV or DVD/Blu-ray. (Yes, you gain some if the final editing /color correction work is done in a 10-bit CODEC). But, remember that all TV and DVD/Blu-ray is 8-bit Long-GOP format. That said, I can understand the need to always capture at the highest possible quality.

I'll have much more to say about this topic in the coming weeks. It's not quite as simple as 8-bit vs 10-bit as many other factors come into play.

Best-

Piotr Wozniacki · June 9th, 2009, 09:59 AM

Understood, Mike and Dan -

I wrote:

Quote:

Originally Posted by Piotr Wozniacki

Dan and Mike,
... If only it's in the realm of the NanoFlash basic design

- didn't I :)

Dan Keaton · June 9th, 2009, 10:28 AM

Dear Piotr,

Yes, you did.

We were just trying to be a clear and informative as possible.

At this time, a very small, low cost, low power, very high quality, 10-bit HD recorder is hard to do.

Mike Marriage · June 9th, 2009, 10:53 AM

Quote:

Originally Posted by Dan Keaton

At this time, a very small, low cost, low power, very high quality, 10-bit HD recorder is hard to do.

Fair enough.

But when it is possible, please can you make one....?

Dan Keaton · June 9th, 2009, 10:56 AM

Dear Mike,

We will be happy to oblige.

Bill Ravens · June 9th, 2009, 11:34 AM

For the time being, I'm content to capture to 8-bit. Provided that I can immediately transcode to 10-bit and edit in 10-bit. Since this is a kinda slow day, here, I'll mention that I'm holding my breath for Cineform to get it together enough to fix NeoHD. I really want a 10-bit pipeline. I've been using Bitjazz's Sheer codec, but, it's not really made for Windows, just yet. Lots of hiccups and playback is stuttery.

Oh well, thanx for the feedback Mike and Dan. You guys are great.

Mike Schell · June 9th, 2009, 07:37 PM

Hi Bill-
Thanks! I'll give the Cineform folks a call and try to accelerate the process. We do get plenty of calls asking for Cineform compatibility.

Best-

John Quick · June 20th, 2009, 11:09 PM

Quote:

Originally Posted by Dan Keaton

Dear Piotr,

We do not know what Vegas does not accept our I-Frame Only.

We are working with Sony on this.

Sony is very cooperative.

Hey, Dan:

Any new word from Sony on this?

Dan Keaton · June 21st, 2009, 05:48 AM

Dear John,

Sorry, but I personally do not have an update on this at this time.

Nolan Craigwell · June 22nd, 2009, 09:40 AM

why can't 220 bs Long Gop be acheived? and i am now seeing that 140 Long GOP has been added. whan can we expect coments on visual differance of 140 vs 100 Long GOP?

Dan Keaton · June 22nd, 2009, 09:52 AM

Dear Nolan,

That is a hard question.

We do not feel that we can reliably create 220 Mbps Long GOP footage, so we do not offer it.

But, there are other considerations.

1. Long GOP is typically 2 to 2.5 times more efficient than I-Frame Only.

2. All that is needed is a compression scheme that gives a recording that is visually indistinguishable from the original and handles motion, and very high detail in the image without any artifacts.

Our 100 Mbps Long GOP does this. And for extra insurance we are offering 140 Mbps. We may add higher bit rates later.

We hope that our users will test these new options and pick whicherver is best for them.

We feel very confidient that 100 Mbps Long GOP is the sweet spot. But, of course, some like to use Intraframe compression, so we offer that also.

June 9th, 2009, 01:46 AM	#16
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear John, The exact performance of the 220 Mbps I-Frame Only is still to be determined. In other words, we do not know which will be better in terms of absolute image quality, ignoring the size of the files. We feel that they will be close, but it is too early to tell. The "Holy Grail" is a codec that produces an image that is visually indistinguishable from uncompressed, and produces the smallest file sizes. Our 100 Mbps Long-GOP mode, in our tests, and in tests performed by others, meets this standard. But, some want an I-Frame Only (Intra-Frame) option, so we offer both. The 100 Mbps option is what I recommend, as this has been proven in many tests, and it produces very compact file sizes (for high quality HD images). The 220 Mbps I-Frame Only file sizes are larger, thus recording time is reduced. __________________ Dan Keaton Augusta Georgia

June 9th, 2009, 09:29 AM	#19
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Bill and Piotr, We fully understand, we would love to do what you want. All MPEG-2 is 8-Bit. All HD Television is 8-Bit. Even if we made a custom version of MPEG-2, the NLE systems would have to support it for it to be really useful. Our Flash XDR will, in the future, with an extra cost upgrade support 10-Bit full uncompressed. This is possible since the Flash XDR has four CompactFlash card slots. We calculate, that using today's CompactFlash card technology, it takes writing to four CompactFlash cards simultaneously to support full uncompressed. The nanoFlash has two CompactFlash card slots. Bill and Piotr, I understand that you know the next point, but some don't: For us to effectively record 10-Bit, a 10-Bit camera needs to be used. There are some 10-Bit cameras at the ultra high-end, and some low cost cameras have 10-bit, but most are only 8-Bit. Just as soon as we are able to provide a 10-Bit solution, we will do so. Richard Welnowski has performed the following test: 1. Record using a great 10-Bit camera, the Thomson Viper, in full 10-Bit 4:4:4. 2, Record simultaneously to our Flash XDR at 100 Mbps Long GOP. 3. Compare each image, frame by frame. The best way I know of to do this is a "Difference Mask" in Photoshop. The results were that the differences were very minor. Some are concerned about banding. We have not seen an instance of banding when using our 100 Mbps Long GOP mode. Most attribute banding to 8-Bit, but it can also be created when using a lower performance codec. No one who has ever tested the Flash XDR or the nanoFlash has ever complained about the image quality or reported banding. I hope this helps. __________________ Dan Keaton Augusta Georgia

June 9th, 2009, 10:28 AM	#22
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Piotr, Yes, you did. We were just trying to be a clear and informative as possible. At this time, a very small, low cost, low power, very high quality, 10-bit HD recorder is hard to do. __________________ Dan Keaton Augusta Georgia

June 9th, 2009, 10:56 AM	#24
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Mike, We will be happy to oblige. __________________ Dan Keaton Augusta Georgia

June 9th, 2009, 07:37 PM	#26
Mike Schell Convergent Design Join Date: Apr 2005 Location: Colorado Springs, CO Posts: 869	Hi Bill- Thanks! I'll give the Cineform folks a call and try to accelerate the process. We do get plenty of calls asking for Cineform compatibility. Best- __________________ Mike Schell Convergent Design

June 9th, 2009, 06:28 AM	#17
Bill Ravens Join Date: Nov 2001 Location: santa fe, nm Posts: 3,264 Images: 10	Hi Dan... I think we've been down this road before. But, at the risk of being redundant, I'd much rather see 10-bit encoding than 220mbps I frame.

June 9th, 2009, 11:34 AM	#25
Bill Ravens Join Date: Nov 2001 Location: santa fe, nm Posts: 3,264 Images: 10	For the time being, I'm content to capture to 8-bit. Provided that I can immediately transcode to 10-bit and edit in 10-bit. Since this is a kinda slow day, here, I'll mention that I'm holding my breath for Cineform to get it together enough to fix NeoHD. I really want a 10-bit pipeline. I've been using Bitjazz's Sheer codec, but, it's not really made for Windows, just yet. Lots of hiccups and playback is stuttery. Oh well, thanx for the feedback Mike and Dan. You guys are great.

June 21st, 2009, 05:48 AM	#28
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear John, Sorry, but I personally do not have an update on this at this time. __________________ Dan Keaton Augusta Georgia

June 22nd, 2009, 09:40 AM	#29
Nolan Craigwell Tourist Join Date: Nov 2008 Location: Trinidad Posts: 3	220 bs Long Gop why can't 220 bs Long Gop be acheived? and i am now seeing that 140 Long GOP has been added. whan can we expect coments on visual differance of 140 vs 100 Long GOP?

June 22nd, 2009, 09:52 AM	#30
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Nolan, That is a hard question. We do not feel that we can reliably create 220 Mbps Long GOP footage, so we do not offer it. But, there are other considerations. 1. Long GOP is typically 2 to 2.5 times more efficient than I-Frame Only. 2. All that is needed is a compression scheme that gives a recording that is visually indistinguishable from the original and handles motion, and very high detail in the image without any artifacts. Our 100 Mbps Long GOP does this. And for extra insurance we are offering 140 Mbps. We may add higher bit rates later. We hope that our users will test these new options and pick whicherver is best for them. We feel very confidient that 100 Mbps Long GOP is the sweet spot. But, of course, some like to use Intraframe compression, so we offer that also. __________________ Dan Keaton Augusta Georgia