nanoFlash Public Beta 1.6.226 Firmware Comments

[Joe Batt]

Billy is the first person to make me laugh out loud at this website. thank you
"thread that refuses to die" HAHA

[Piotr Wozniacki]

Originally Posted by Billy Steinberg

What on earth does this have to do with the new firmware? Why didn't you just include it in the thread that refuses to die ("Noise comparison: 35/4:2:0 vs. 180/4:2:2"); you know, the one where no one is willing to make the only realistic evaluation, where both codecs are compared to an uncompressed frame?

Why are you surprised that there is a difference between the two codecs? And if you felt they should be the same, why did you bother getting a nano in the first place?

Who cares which scope picture is which, and why didn't you include a picture of each of the frames you captured, which would at least make it a reasonable question? All the scope pictures tell us is that the frames aren't the same, not which one is more accurate.

Billy

It's funny how some (most ?) people will deny hard facts in order to rationalize their investment. You have no idea how many of them PM'ed me with exactly the same observations as mine in the "thread that refuses to die" !

I stated it many times the nanoFlash is a great device, and I'm using it every day - yet, it doesn't mean it's forbidden to speak about its shortcomings. Or simply wonder about some of its features, not even calling them "shortcomings".

But enough's been said; I guess I owe an answer to those who tried to guess which scopes belong to the nanoFlash: it indeed is the "scopes2.jpg" picture, so Peter Moretti had it right.

As you say, Peter, "it has a wider vectorscope trace which [...] is a result of having more chroma info". What caught my attention though is that - at the same time - it seems to have less black information (when I compare my other clips, the nanoFlash versions have consistently narrower histogram than their EX counterparts).

Can anybody explain why the nanoFlash would change levels?

[Dan Keaton]

Dear Piotr,

Please post the original frames. We need these to evaluate the scopes.

[Piotr Wozniacki]

Dear Dan,

As I said in my previous post: when I compare my other clips, the nanoFlash versions have consistently narrower histogram than their EX counterparts. Aso, the same can be seen with other nanoFlash formats (like 100 Mbps L-GoP), and with the previous nF firmware.

So there is no point in posting any particular pair of stills (obviously, people on this forum have had enough of them :)). If however my observation is something new to you, and you cannot re-create it yourself - please let me know, and I'll send you a couple of samples with an email.

Thanks,

Piotr

[Piotr Wozniacki]

OK, here is the most neutral comparison:

The difference is extremely subtle, but still the nano histogram is narrower (at both ends - not just blacks)...

It's interesting to see how the single dot in the center of the nano's vectorscope, gets replaced by a circular array of dots for the EX. This might be the reason for "more information" at left side of the EX histogram - if it's indeed, that the "extra information" would probably be just the EX's lower color resolution garbage.

[Dan Keaton]

Dear Friends,

Disclaimer: While I understand a Vectorscope and Histogram, I do not consider myself an expert in using these tools.

With that said, it appears to me that the Vectorscope tells the story.

In Image 3 (nanoFlash), the scopes for the nanoFlash, take a close look at the blacks (center bullseye).
Note that these is only one dot in the nanoFlash image.

Compare this to Image 4 (EX), there is not just one center dot, but quite a few in the center (black) area.

In my opinon, the nanoFlash recording is cleaner and this is easily seen in the Vectorscope.

Besides the blacks, the cyan area of the EX has extra dots that are not present in the nanoFlash image.
And note that some are outside the box.

The histogram for the EX will be "Wider" using Priotr's term, for the EX.
The Vectorscope shows why: there is more noise in the EX 35 Mbps (or at least it is less precise).


My comments are not to trash the EX, but to explain why there is a difference in the histogram,
in other words, to answer Piotr questions.

I welcome comments and suggestions from those that are experts in reading these tools.

[Piotr Wozniacki]

Dear Dan,

As I already stated in my last post, this is what I'm thinking as well.

Thank you for inviting me to post some screen grabs; while I was reluctant at the beginning (for obvious reasons :)), this led me to the idea of shooting the EX1's own bars. Their scopes, and as you point out - particularly the vectorscope, says a lot.

Since I'm not an expert in interpreting all the intricacies scopes show, I'd once again like to invite those more knowledgeable than myself, to chime in and discuss the matter further.

Originally Posted by Dan Keaton

The histogram for the EX will be "Wider" using Priotr's term, for the EX.
The Vectorscope shows why: there is more noise in the EX 35 Mbps (or at least it is less precise).
.

I'd say it's not more noise (in the common meaning of the term), but certainly less precision of the EX 4:2:0 picture when compared to the nanoFlash'es 4:2:2.

To Chris: if you think this subject deserves it, please take out posts starting with #34 from the Beta firmware discussion thread and create a separate one. Thank you.

Piotr

[Dan Keaton]

Dear Piotr,

After I posted my reply, I re-read your original post and noted that you came to a similar conclusion.

[Piotr Wozniacki]

Dear Dan,

Yes - this is the only explanation that comes to my mind.

This can be further illustrated with the examples below: on the right, you can see a blown-up detail from the EX. At 4:2:0 color resolution, you can see black stripes across the solid red color - they may be accounted for by the extra content at the very left end of the histogram.

The left-hand side pics show the same frame from the nano; no black stripes, less black garbage in the histogram, richer vectorscope trace...

[Peter Moretti]

I believe the six vectorscope dots around the center are caused by bleeding between the luma and chroma values along the horizontal edge where the vertical color bars meet the grey horizontal stripe. This is causing values that are "mostly white." That is, they have almost equal R, G and B vlaues but do also contain small amounts of color information from the color bars.

Such values would show as being slightly off center in the direction of the "polluting color" on the vectorscope's graticle. And that's exactly what we see.

If you blowup by a considerable amount the border between the horizontal line and the color bars, you can actaully see this "bleeding" with your naked eye.

I HTH and believe it's accurate. ;).

[Piotr Wozniacki]

Originally Posted by Peter Moretti

If you blowup by a considerable amount the border between the horizontal line and the color bars, you can actaully see this "bleeding" with your naked eye.

Yes Peter - the vertical color resolution is so much better with 4:2:2 than it is with 4:2:0, isn't it. It can be seen at the border between the horizontal line and the color bars; it's also apparent in my red flower blow-up.

How nice it would be if the same was possible in the other direction - but there is only one way to get it all: 4:4:4 :(

[Peter Moretti]

Actually, 4:2:2 and 4:2:0 theoretically have the same vertical resolution. It's the horizontal resolution that's different.

But when you throw in macro blocking, bit rate and how the codec works, there can be a difference in vertical resolution as well. So the horizontal banding you're seeing is not really the direct result of 4:2:0 but a more complex combination of factors.

[Dan Keaton]

Dear Friends,

I received an email concerning these comparisons:

He points out that "the difference between the SxS and the NANO files, is not due to the difference on the recorded signal, but on how is displayed."

"The difference is dues to a property of the QT files called APERTURE ( Properties > Presentation).

The aperture of the NANO files is "CLASSIC", while the aperture of the SxS files is PRODUCTION.

QT has four different aperture options.

The same clip displayed with different "apertures" will show some differences on the screen (QT Player, FC) and on the VideoScope."

[Peter Moretti]

Dan,

I respectfully disagree w/ that explanation. The aperture size in QT has to do with subtle differences between SD and DV frame sizes and pixel aspect ratio. It would not cause the horizontal banding that I described above.

A difference in QT Aperture can cause a difference in adding or removing black from the the borders of the entire frame, but I really don't think that's happening in this case.

HTH, and that I'm giving you the correct info.. (I obviously believe that I am.)

[Piotr Wozniacki]

Originally Posted by Peter Moretti

Actually, 4:2:2 and 4:2:0 theoretically have the same vertical resolution. It's the horizontal resolution that's different.

I beg to disagree - just take a look at diagrams on Page 3 here:

http://dougkerr.net/pumpkin/articles/Subsampling.pdf

But I guess we mean the same thing - just a matter of nomenclature; better vertical resolution = more horizontal lines.