Does the nano help reduce interline twitter on EX-1

[Bill Ravens]

I think you’ll find the twitter problem is all but eliminated at bit rates of 100mbps or higher. As an aside, are you using much sharpening in your camera setup? Sharpening, in-camera, exacerbates the twitter problem.

[Rafael Amador]

Hi Willians,
How the NANO improve the pictures on pans you can see it at 50Mbps.
About the INTERLINE TWITTER, that’s not a matter of the NANO.
The NANO just record what the camera outs; If the INTERLINE TWITTER is the SDI you won’t get rid of it even if recording fully uncompress.
Adjust the camera.

[William Urschel] “I am sitting here already with $5,000 worth of SxS cards (purchased at that full price 1 1/2 years ago”
This is why last year, instead of buying more SxS cards, I bought the NANO.
For 5k you can buy the NANO and 2.000 $ worth of CF cards; that means some 32 hours EX recording (when buying top quality CF cards).
Sell the cards and buy the NANO; even if you don’t care of the better color space.
Best,
Rafael

http://www.nagavideo.com

[William Urschel]

Thanks, Bill (Ravens)! Am NOT using in camera sharpening – spent 1,700 hours (!) last year, after hours, according to my time reports, doing everything I could think of to get rid of BD exhibited interline twitter and abomination of down rezzed DVDs. Finally gave up on DVDs entirely, except the beauties produced in downrezzing those resulting from HD on all my Sony HDV cameras!

Amoung my “experiements”, under very carefully controlled and documented conditions: 1)tried 5 different settings in PP for “detail” on the camera – finally removed twitter, but the pictures were so badly smeared that one of my old, 1991 Sony cameras
produced far, far better pictures (in SD, of course). And I purchased Vegas Pro (to try out versus Adobe Production Permium CS-3 and CS-4), and SEVEN other apps (darn, some were quite expensive!) for downrezzing from HD to SD and encoding for BD and DVD, all to NO avail. I have been using Cineform (last, Prospect 4k) for years, and to GREAT advantage in handling HDV editing enhancement and backup. But, absolutely surprising to me, going to Premier, Adobe Encoder, and Encore in CS-4, without Cineform, produced the least artifacting and interline twitter.

And by the way, I have had other editors check out my work flow, etc. to no avail.

I have come so close to getting rid of the EX-1, until I went to CS-4 without Cineform (Cineform wil not work with appropriate functionality on Premier Pro CS-4, though it will probably come back with CS-5!)- I’m still getting some interline twitter, but nothing like what I had before, and I am now partly satisfied with the results.

In any event, Bill, your comment about going to 100k with nano as a method of avoiding interline twitter – QUESTION – is this opinion based upon your or one of your compatriot’s actual experience? If so, then I don’t care what it costs, I’ll go for it, to end over a year of total grief and frustration for this otherwise absolutely superb camera!

Thanking you or any others for your informed response, I remain your humble servant,

Bill Urshel

[William Urschel]

Rafael

Thank you – good point re selling! Probably get less than $400 a card (very poor return on investment!), but every little bit helps, doesn’t it!

Thanks again.

Bill Urschel

[Michael Slowe]

Having also been driven mad by line twitter on SD DVD’s (don’t see it on tape) i found that the new versions of the MPEG 2 encoder BitVice virtually solved the problem. It’s a great encoder and Innobits (the Swedish company who make BitVice) are continually developing and improving their product. The problem is of course that the EX cameras are giving us too sharp an image!

Michael Slowe

[Tim Kolb]

Hmmm… several issues here.

“Interline twitter”…I’ll assume we’re referring to what engineers would refer to ‘aliasing’ or ‘mosaic’ which would occur on clothing with a fine herring bone pattern where it looks as if the cloth is crawling off the person…and is most common in interlace signals.

If you are shooting 30p, I’m not certain how you would produce this artifact…at least to the extent you would see it in interlaced mode. The monitor you use to view your video has an impact on what you see vs what you actually have…not to mention that on many NLEs, the viewing window (panel, canvas, whatever) will usually scale the actual pixel size to fit within the interface and this will also produce some interesting artifacts that aren’t actually in the file.

Your HDV gear (if it’s Sony) is likely dropping to one field if you are choosing progressive mode, effectively halving the vertical resolution (to 540 vs 1080), softening high frequency artifacts far more than an EX1 would, which is shooting full resolution vertically (1080), not to mention 33% more resolution horizontally (1440 vs 1920).

As far as panning goes…with progressive scan, you will never have the same smooth motion as you would with interlace. This is the reason why cinematographers have very specific speeds that they know they can pan with any given lense…they have to adjust their technique to reduce “judder”.

100 Mbit I frame vs 35 Mbit Long GOP… I’m an advocate for the Convergent Design product line as much as anyone. There is no question that the Flash XDR and the Flash Nano are really well thought-out devices and that 100 Mbit I frame is very good quality.

However…there are a few caveats here.

1. Temporal compression is typically AT LEAST 2.5 times as efficient as I-frame only codecs, all other compression traits being equal. This being the case, 100 Mbit I-frame MPEG would carry about the same image quality as 40 Mbit long GOP. Again, this is assuming all else is equal. So going to 100 Mbit I-frame likely won’t grab the advantage over 35 Mbit Long GOP that one would assume by judging on the raw numbers… (100 Mbit Long GOP 4:2:2 is an obvious jump however).

2. However…XD(EX)cam is 35 Mbit 4:2:0 whereas the Nano records 4:2:2 at data rates above 35 Mbits, so you have denser color difference subsampling. Advantage for quality goes to the Nano on this point…

3. BUT…whether this would resolve the fine pattern noise issue is in question in my mind. The camera’s low pass filter system combined with it’s detail circuitry is where the rubber hits the road here…and the Nano and the camera have equal recorded luma resolution…1920×1080. I don’t think the fine pattern noise will be affected by the recording system in this case quite frankly.

I’d say the Nano is a very sound investment…but I think that the nature of the visual issues you are referring to lead me to believe that there is more to this story and that without some other alteration in how you are using the camera or you are viewing the footage, I’d be surprised if switching to the Nano -alone- would solve the problem.

…I will say that my conclusions are based on what I believe you are describing. If I’m misinterpreting your comments, feel free to clarify.

TimK,
Director, Consultant
Kolb Productions,

[Bill Ravens]

I agree with you Tim. Your post made me think of another factor. Downrezzing HD(1920×1080) to SD(720×480) is inherently a problem, quite different from the twitter that results from bandwidth limiting and compression artifacts. In my experience, if one is aware of the need for downrezzing in their workflow and delivery, 1280×720 makes a much prettier source for downrezzing. Having said this, there are several techniques posted on the internet for mitigating line twitter when converting HD to SD.

[Don Greening]

Great post, TIm. It’s worth a light bulb.

– Don

[Rafael Amador]

[Tim Kolb] ” which would occur on clothing with a fine herring bone pattern where it looks as if the cloth is crawling off the person.”
Tim,
Where there is a lot of B&W detail, the frequency of the Y channel some time gets close to the frequency of the Color Subcarrier. The filters read that Luma as Chroma,producing these ever changing colors.
Is not a fault of the camera, but of how the NTSC and PAL system have been designed.
That’s call “moire” and there is nothing to do avoid it.
I guess that you can tweak the detail to avoid it, but probably will kill the picture.
Best,
Rafael

http://www.nagavideo.com

[Tim Kolb]

[Rafael Amador] “Where there is a lot of B&W detail, the frequency of the Y channel some time gets close to the frequency of the Color Subcarrier. The filters read that Luma as Chroma,producing these ever changing colors.
Is not a fault of the camera, but of how the NTSC and PAL system have been designed.
That’s call “moire” and there is nothing to do avoid it.
I guess that you can tweak the detail to avoid it, but probably will kill the picture.”

Well…aliasing happens in many instances and from many sources and NTSC color isn’t the sole cause…subcarrier moire is one…though in HD, there is no analog color subcarrier.

If you have any high frequency detail that is above 50% of the max sensor spatial resolution, aliasing will happen. Nyquist tells us that you need to sample at 2X the max frequency you wish to faithfully reproduce. This is the reason why the low pass filter on any electronic camera should be cutting out detail above 50% of the max sensor resolution, but there are still situations where aliasing can happen.

Turn up the detail on your video cam and find out…

TimK,
Director, Consultant
Kolb Productions,