DV Magazine Review of the nanoFlash

[Rafael Amador]

What Discovery Channel have done is to demonstrate that with low prices cameras and a very good CC and postproduction, they can save lot of money and the audience doesn't goes down.

I guess they use the best cameras on the world when needed (they can afford it), but, for much of what they show, made of shots where the eye have no time to repair in details, these cameras works perfect.

The video technology is founded in the limitations of the human eye. Why to give him more than he can takes?
rafael

[Alister Chapman]

Sorry Mark, I did misunderstand you. You did say HDV fests wired back to VTR's, which I took to mean HDV cams wired to VTR's.

I know about their use of POV cams, which are not HDV cameras at all and only make up a small part of the programs content. I've spent many days working with Reed, Josh and Tim and supplied lots of footage for the previous 2 series. Yes they use HDV but to be honest the show is hardly a quality benchmark. When I have suppied clips to them I have had to take my XDCAM HD422 material and dub it to HDV, shame!

What I would consider quality benchmarks are the BBC Natural History Unit programs, shows such as Planet Earth or Life. Even today these are predominantly shot with the old Panasonic HDC27 Varicams, using DVCPRO HD... which is 8 bit... and look pretty good to me.

[Mark Job]

Hi Alister:

You wrote: "Sorry Mark, I did misunderstand you. You did say HDV fests wired back to VTR's, which I took to mean HDV cams wired to VTR's."

....No problem Alister. Do you know if Sean Casey, the man who built the TIV, has completed his IMAX film on tornadoes yet ? Last I heard, he was having great difficulty with some of the chasing community, as they have come to now be called. Planet Earth was shown twice here in Canada on CBC. The footage from that series looked fantastic.

[Alister Chapman]

No he has not finished it as he has yet to get that elusive shot from inside a daytime tornado. Sean gets a lot of flack from many chasers as he does some pretty crazy things that give storm chasing a bad rap, things like going the wrong way down interstates or totally blocking roads. I'll never forget when he used to follow us in a pickup truck with his IMAX camera on a tyler mount in the truck bed. Watching him standing in the bed, swapping 75mm film mags with strips of film flapping everywhere while driving down I40 at 80mph was a sight to behold!

[Olakunle Olanrewaju]

Originally Posted by Mike Schell

Just an FYI, we are planning to add a 10-Bit to 8-Bit dither to the nano/XDR firmware early next year. We think this should largely eliminate any potential banding issues associated with 8-bit processing. Naturally this requires an HD-SDI source with true 10-bit resolution.

Best-

Pardon my ignorance on 10-Bit to 8-Bit dither. does it mean it allow recording in 10 bit 4.2.2 and when is likely to come out this year.

[Adam Stanislav]

Originally Posted by Olakunle Olanrewaju

Pardon my ignorance on 10-Bit to 8-Bit dither. does it mean it allow recording in 10 bit 4.2.2 and when is likely to come out this year.

No, that's not what dithering is. It is a method of error diffusion, but it will still be recorded in 8 bits.

At any rate, I hope it will be optional, i.e., we will be able to turn it off. Personally, I hate dithering of digital images.

[Mark Job]

Hi Adam & Olekunie:
The only way to get a true 10 bit signal into a Flash XDR is to enable an uncompressed recording option. Such a measure would allow the signal to pass directly from an HD-SDI 10 bit input source, then by-passing the built in hardware Sony XDCAM HD 4:2;2 encoder, thus no encoding or dithering would be required. In this manner the signal is directly written back out to the CF card media. Such an option is mainly advantageous for digital cinema origination where video must be output to 35 MM film for standard cinema distribution. You would have to be able to stripe 4 x 32 GB or 64 GB CF cards as a RAID 0 volume to insure adequate data transfer rates. You could have two output options in such a scenario with the Flash XDR.

1. You play out of the Flash XDR in realtime via the HD-SDI outputs.
2. You enable the IEEE 1394 interface on the XDR and transfer your data to computer NLE. The transfer would be long with 4 x 64 GB cards, but quite manageable with 4 x 32 GB cards.

To my knowledge, the Nano Flash does not have the internal hardware capability to have this option, but the Flash XDR does.

[Olakunle Olanrewaju]

Thanks for the explanation but from all indication it appears that the flash XDR has been discontinued

[Mark Job]

Originally Posted by Olakunle Olanrewaju

Thanks for the explanation but from all indication it appears that the flash XDR has been discontinued

....Hi Olakunie: Yes. This is correct. THe Flash XDR is no longer being manufactured but is still being supported by Convergent Technologies.

[Tim Kolb]

Originally Posted by Adam Stanislav

No, that's not what dithering is. It is a method of error diffusion, but it will still be recorded in 8 bits.

At any rate, I hope it will be optional, i.e., we will be able to turn it off. Personally, I hate dithering of digital images.

Dithering, in general, is a method of creating areas of pixels with multiple values in a pattern to visually mimic some value otherwise unachievable resulting from the combination of colors "mixing" in our vision...

Taking a digital photograph and using Photoshop to create a lower bit depth image that would be as faithful to the original as possible would likely employ dithering.

In the past, dithering has been evidence that our computers don't have the juice to show us precise enough color depth to carry honest image detail...about as welcome as DCT macroblocking in CinePak video clips. (Now, I'm dating myself).

However, in the present day and working with HD and larger images, it is common practice in VFX and color correction workflows to combat banding issues inherent with 'pushing' 8 bit images aggressively (or even not so aggressively), to add some subtle element of noise to the image. It's an illusion of course, but many respected Post/VFX types claim to be able to gain what they consider the visual perception equivalent of an extra bit of precision (8 bit appears that it could be 9 bit, etc).


As far as the review that started all this... I don't think that Ned was implying that 8 bit was a deal-breaker, but if there are 10 bit alternatives available (the KiPro is always the one brought up) that do 10 bit...I'm not sure how you can say 8 bit is -better- than 10 bit... it's just one of the parameters. If you're a reader of the review and this isn't a big deal...there is no problem. He could have said that it's a disadvantage that the device doesn't mount directly to a specific apparatus on an Alexa. I guess an Alexa operator who wanted a NanoFlash on board for dailies might have to consider how to mount it (and would likely have anything they needed to deal with it.). For me...a non-issue.

This 8 bit vs 10 bit discussion is largely academic for the majority (other than high-end entertainment work) of present-day, non-VFX image acquisition, I think. Television shows used to be the technological gold-standard we all aspired to... Now I think that television shows run the gamut from extravagant, uncompromising quality to experimental ("House" on DSLR) to just plain expedient (Deadliest Catch, etc). No one in their right mind would shoot "Deadliest Catch" with a half-dozen F35s. That would be stupidity. Shooting "Desperate Housewives" on a Z1 would be equally so...

Is 10 bit acquisition helpful? Tough to say no... Is 10 bit necessary? Obviously that isn't true in every case... AVC Intra has this capability, but I would think the difference from 8 bit would be far more meaningful on Panasonic's higher end cameras than it would be on the lower price point camcorders. ProRes is just simply tailor-made for FCP and it streamlines the post process, and I think in many cases those who would choose a KiPro do so for that convenience and consider the 10 bit precision to be an added bonus or validation. I suspect that if the Nano acquired a 10 bit file that FCP had to log and rewrap, and the KiPro captured ProRes only at 8 bit (for argument here), i still think that many who are looking at their FCP workflow might still choose a KiPro.

I use them both. I don't see either one as being able to replace the other actually. I suppose that Convergent could come out with something that looked a bit more like a VTR with similar controls and lots of CF slots for a massive record load, but I think what they're focusing on now is what they're doing well. I don't want a studio VTR replacement 'FlashJumbo'.


If reviewers are to help us understand how products in the marketplace relate to each other, you have to be able to point out what each device has and doesn't have, does and doesn't do, and where it seems to fit in the overall landscape, otherwise they do us no service.

[Adam Stanislav]

Originally Posted by Tim Kolb

I'm not sure how you can say 8 bit is -better- than 10 bit...

I didn't say any such thing. Nor did I imply it.

[Tim Kolb]

Originally Posted by Tim Kolb View Post
"I'm not sure how you can say 8 bit is -better- than 10 bit..."

Originally Posted by Adam Stanislav

I didn't say any such thing. Nor did I imply it.

I wasn't implying that you said this...I was thinking out loud in my response to those who were critical of the reviewer for listing 8 bit as a drawback...think of me scratching my chin and questioning how else the guy could have addressed it in the article as opposed to challenging anyone...

Hey...we're from Wisconsin. We're always nice.

:-)

[Adam Stanislav]

Yes, we're always nice. :)

[Mark Job]

Hi Friends:
Actually 8 bit can be very good. My Canon XL H1 is an 8 bit camera (At least it is when you go through its internal HDV 4:2:0 encoder and out to its built in VTR). What gets confusing with the XL H1 is it also has a full raster HD-SDI *Uncompressed* output. Now, my understanding of this is uncompressed Full Raster HD is inherently 10 bit (?) !! So what is the H1 camera doing exactly to give the full raster in an uncompressed form from what is normally 1440 x 1080 in 8 bits ?????????? I'm sure the H1 uses Anamorphic pixelation to do some kind of....What ?....Exactly.....Is it doing ??????

1. Is it stretching the raster ?

2. Is it scaling the raster ?

3. Is it dithering its native 8 bit signal up to 10 bits ???

*This kind of confuses the heck out of me ! I wish we could have a genuine Canon Technician who designed the internal camera signal architecture come on this thread and explain how it works and also the merits of 8 bits vs 10.

I do know I can see one heck of a difference with primary and secondary color correction when I do a video mixdown of 8 bit HDV to 10 bit Sony XDCAM HD or the Avid DNxHD 220 Mbps 10 bit codec and do all my CC in 10 bit ! Otherwise I'm posting Ep2 of my online web series in DNxHD 145 8 bit and it looks great ! There is no banding I can see.. The Flash XDR and the Nano Flash are 8 bit recorders. The video out of them looks fantastic !

[John Mitchell]

I think the answer is Mark that CD throw away the extra 2 bits.

The advantage with dithering would be that some kind of algorithm would attempt to round those extra 2 bits into the first 8 bit number giving a smoother transition. If you think about it in decimal terms (which is a clumsy analogy) if you had a number like 1.000000094 and you wanted to "dither" that number by two decimal places you would end up with 1.0000001. Currently the Nano wold just throw away those last two numbers and end up with a slightly different result...1.0000000.

I'm not sure what the impact on the final image would be - maybe slightly smoother gradients?