...but 220 Mbps I-Fo is our new, true "sweet point" of nanoFlash recording.

Is this what you mean, Gints? If so, I'm with you on this.

Especially now, when the new Vegas Pro 10 has been improved sooo much in the high bitrate nanoFlash file decoding-playback (even the 280 Mbps clips play full fps, with the full quality - something impossible in the previous version, where 100 Mbps L-GoP was the highest bitrate format that played back without stuttering).

Piotr

"but 220 Mbps I-Fo is our new, true "sweet point" of nanoFlash recording."

I used DV25 to show that 220 Mbps I-Frame is not that crazy for HD and that we shouldn't expect great compression quality from I-Frame at lower bitrates. If anything, 280 Mbps does not look as crazy as before.

Hi Gints, Piotr, and Andrew:
The bad news, is 8 bits isn't quite enough, but the good news, is it certainly can be "finessed" and manipulated into looking much better than it would normally. I find in many NLE post workflows, if you fail to dig deep down into your project settings to make sure you have set the highest processing pixel precision possible, then you wind up "coloring" your video image quality in a degrading way. If you don't set the audio project bit depth to at least 24 bits (Regardless of source audio bit depth ), then you also degrade the overall audio as soon as you add filters and begin mixing and rendering the work down. What Piotr and Gints have seen, is not so much the deficiencies of the Nano, rather, it is the overall deficiencies of the MPEG 2 Long GOP specification, which will repeatedly reveal its limits on large screen TV's if the source footage is not processed adequately in the post workflow.
Now with these new super resolution camcorders coming down the pipe, these subtle differences will only become even more apparent visually. The subtle detail, which an ill adjusted NLE will lose for you, is what makes the picture when its on a 52 inch plus size screen. The big screen TV's are so good these days they are very capable of reproducing all noise and coding inefficiencies of 8 bit MPEG - 2, unless you finesse the heck out of it.
Ask yourself the basic question - How do you use your SSDR ? In 98 % of the cases I work on, I use my Flash XDR as a simple black box recorder, this is to say, I only use the box to capture and playout the recorded clips from the XDR itself, or I import the clips directly off of the CF Card media to edit. I do not use, nor do I need any of the other features - except the odd time-lapse sequence, but now even this has been done to death in television. I am of the minority opinion on this forum when it comes to how I view the role of the Solid State Digital Recorder to be.

** I see the role of the SSDR to be able to capture what the camera **sensor** can actually see, and put this data onto solid state recordable media in the best possible quality, with little or no compression in as high a color space and pixel precision as possible. I consider *Compression* to be the role of the NLE - NOT the SSDR ! You should compress your data at the NLE stage, if any, and leave the field recorder to treat the source image data as "digital film." I'm designing my long and painful SSDR project to do just that. These new Panasonic & Sony Cameras are pretty exciting, and I believe the option of uncompressed capture will eventually make its way down to the lower end consumer recorders due to the ever improving media density vs price point in removable storage, plus the superior signal to noise ratio of the video response signals coming out of these new digital cinema like camcorders.

Agreed - at anything lower than 180 Mbps (an absolute minimum), I-Fo looks like crap. Much worse than L-GoP at corresponding bitrates.

Piotr

Dear Piotr,

We do not recommend shooting at 100 Mbps I-Frame Only.

I mention this to our customers all of the time. This is not a recommended option.

Sure thing Dan - on this particular thing, there has always been a consensus between us :)

Piotr

Final Cut renders in 32b Floating point when needed.
Just set "Render all YUV in High Precision".
Rafael

Just as I said re: Vegas Pro.

Agreed - at anything lower than 180 Mbps (an absolute minimum), I-Fo looks like crap. Much worse than L-GoP at corresponding bitrates.

Piotr
As expected.
Piotr, think that the NANO at 280Mbps Intra, is virtually 8b Uncompress.
You get a file just a bit bigger just by Zipping.
rafael

Dear Dan,
All the pals around, and my self, would be very happy to work with Convergent-Design on trying to fine tune the NANO's LGOP structure.
There is nothing to be done on the Intraframe but rising the data-rate (no needed IMO).
But about the GOPs, I think that, together, we can make really shine the SONY processor.
I have two months free to try any Beta.
Rafael

Well, it's been a long time since the last post in this thread...

Of course in the meantime Convergent Design has been busy with their revolutionary Gemini device. I sincerely do hope that we - the nanoFlash users, and minor players (when compared to those being the target for the Gemini) - are not left alone, with some obvious shortcomings of the current firmware (like no cranking to MXF format), and unfulfilled promises (like mixing the SDI/HDMI embedded audio with analog channels).

But even if CD still does care for us, I guess any hopes for fine-tuning the Long-GoP encoding should be buried now.

Nevertheless, I just feel obliged to inform you guys that when used to record the super-clean image from the FS100 camera, my nanoFlash really is showing its great potential:

- with virtually no noise present in the (HDMI, in this case) signal, both the L-GoP and I-Fo clips are absolutely gorgeous. Even if one can find some macroblocking or other codec-stressing artifacts in the FS100's native AVCHD clips, they are not present in the same material recorded simultaneously to the nanoFlash!

If only we had full HDMI/TC support ....

Piotr

"I guess any hopes for fine-tuning the Long-GoP encoding should be buried now."

This may not be possible without increasing the data rate. At least on the Nano, the I-frame appears to improve significantly with increases in data rate at 100 Mbps and higher while the other frames do not increase in quality as much. "Constant Quality" may not be implemented or even possible in current real-time Sony compression systems.

The solution for the Nanoflash is to buy 6x more flash and record at 280 Mbps I-Frame only. It's just too bad that this option is so costly even in 2011. It would be better to have a SATA interface for SSD and spend $500-1000 on quality 256GB SSDs than $1000/128 GB of SanDisk Extreme flash.

You are right Gints.

I can confirm that while the L-GoP does not seem to increase noise with increased bitrate (which is occurring with a noisier sources like the EX1), I-Fo is still my format of choice - also with FS100. However, 220 Mbps looks like the sweet spot for me. Don't see a real need to go any higher.

Piotr