Cinematography Mailing List - CML
    advanced

Aliasing On Right-hand Side Of F900

 

Hello,

Just wondering for any other F900 owners out there, I'm getting this weird aliasing artifact on the right edge of the frame of an F900 I've been using for the two shoots.

An example of it is here at :

http://home.mindspring.com/~jrod/Aliasing.jpg

I thought at first it might be the lens, but it's on a couple different lens combinations, so that rules that out. I'm not sure if this is something common to all F900's, or if I should call the rental company to see if they need to service the camera.

Thanks,

Jason Rodriguez
Post Production Artist
Virginia Beach, VA


>Just wondering for any other F900 owners out there, I'm getting this >weird aliasing artifact on the right edge of the frame of an F900 I've >been using for the two shoots.

---That doesn’t look like aliasing...it looks like your camera is out of registration.

Jeffery Haas
freelance editor, camera operator
Dallas, Texas


Jason Rodriguez wrote :

>I'm not sure if this is something common to all F900's

I think you will find that it is. I have heard several explanations for it, the latest being the prism.

Noel Sterrett
Admit One Pictures
www.admitone.com


Jason

It looks to me like there is definitely a problem with the camera. This is NOT normal and really does not look like what I have seen that Noel speaks about, and NOT a registration error. It appears that it might be an A>D converter or DSP malfunction. Do you see it at other frame rates? Is this off tape in the camera?

Michael Bravin
Chief Technology Officer
Band Pro Film & Digital
www.digizoom.com


Yes, It's off tape, and it's happening from pixel 1907-1920 horizontally, so around the right ten pixels of the screen.

I guess I'll call the shop and have them take a look at the camera.

Jason Rodriguez
Post Production Artist
Virginia Beach, VA


>It's off tape, and it's happening from pixel 1907-1920 horizontally, so >around the right ten pixels of the screen.

I'm not sure I see what you are calling aliasing. I've seen HDCAM "mangle" roughly 10 to 16 pixels on the left and right of each line due to filtering (most recently during testing at the BBC).

You don't even have to record the image to see this, just E to E.

Martin Euredjian
eCinema Systems, Inc.
www.ecinemasys.com


Hi Martin,

Yes, this seems to be what you're talking about. The images I posted were off HDCAM tape, and as you can see, certain lines along the edges, about 10 pixels from the right-hand side in this case, seemed to get vertically streaked, or what seems like really bad aliasing. It's like the image has been shifted slightly on the vertical every two-three pixels, so it looks like bad aliasing.

Martin, is there a paper or any other example of this that I can find online? The company I've been getting my F900 from has one in-house that they typically give me, so I haven't been able to test out any other F900's, and I probably won't be able to get a hold of another one till the end of August/September if I decide to use one on my next shoot. I don't want to raise a red flag if there's no need to.

Thanks,

Jason Rodriguez
Post Production Artist
Virginia Beach, VA


> It's like the image has been shifted slightly on the vertical

OK, that's not the HDCAM processing problem as far as I've seen it. What I'm talking about is pretty much confined to the horizontal dimension. You need to frame capture directly off the camera in order to avoid introducing HDCAM issues, and, as I said before, just going E-to-E through an HDCAM deck will distort the signal to a significant degree. Bypass the deck entirely.

Also, is this something you are seeing on a monitor or only post-recording/capture? You need to eliminate all non-camera items from the equation. How are you monitoring the output of the camera?

Also, is this a frame from a sequence where there's motion or is it locked-down still without any motion whatsoever?

Martin Euredjian
eCinema Systems, Inc.


Jason :

>Just wondering for any other F900 owners out there, I'm getting this >weird aliasing artifact on the right edge of the frame of an F900 I've >been using for the two shoots.

I wouldn't call it aliasing -- it's more like the right-hand twelve pixels got grouped into three groups of four, and those became super-pixels. Did you count pixels to see if these pixels fall outside of 1920 x 1080? Perhaps they're a weird artifact -- I don't know your signal path, so it's hard to guess.

You are also getting color fringing on most of your edges -- magenta to the left, green to the right. I'm assuming this is optical, but it's sure not pretty. Look around the fluorescent lamp, or around the piece of wood trim on the wall, or nearly anywhere.

While the mangled pixels are certainly a problem to be solved, they are limited to the extreme outer edges of the image. The fringing degrades the entire image.

What lenses were you using? I will assume the back focus was tweaked, and that you had a Zeiss focus sticker on your camera -- as a totem, if nothing else.

Jeff "they clearly weren't Digiprimes, I'm guessing" Kreines


It's a fairly still frame on stix, but it's not locked down, there's motion in the shot itself, and I've measured the pixels, they're within the 1920 boundary-like I mentioned before, they're around pixel 1907 out to the full 1920, so the last 10-13 pixels on the horizontal axis.

I never saw this on the monitor, although we were using the Analog outs on the camera, and the monitor was one of the small portable ones: one shoot was an Astro, the other the Sony 9". I'm not sure you could see the problems on those monitors, at least I never noticed it, but then again I was directing, not really paying attention to artefacts on the side of the frame.

I've only noticed this after seeing the footage full-res after digitizing, which I used a J-H3 deck an a Cinewave system, so you're looking at uncompressed HD over SDI from the J-H3.

It's on a number of different footage shots now that I know what to look for. Also doesn't matter if the shutter is off or on (at first I thought it might be motion blur artifacting). And again, it's also happened with a couple different lenses, so it's not the lens (also my first assumption).

You're right Jeff, definitely not a Zeiss!! The shot I uploaded was a Canon 11x4.7 at around 5mm, and you're looking at the very edge of the frame. Also stupid me had the highlight saturation function "on", so it coloured the highlights magenta! Had to CC all that out. I always keep that "off" now!

Here's another example, this time on non-straight edges (the holes on the back of a monitor) :

http://home.mindspring.com/~jrod/MonitorBack.jpg

(also shot with the 11x4.7, so again lots of CA on the edge-this shot is also in slight motion unlike the other one I uploaded).

Thanks everybody for helping me to track this down,

Jason Rodriguez
Post Production Artist
Virginia Beach, VA


You need to eliminate your recording and digitizing system from the equation. Do you have another source? A test signal generator? How about going into Cinewave directly off the camera? Have you done that?

Martin Euredjian
eCinema Systems, Inc.


Jason Rodriguez wrote:

>I've only noticed this after seeing the footage full-res after digitizing, >which I used a J-H3 deck an a Cinewave system, so you're looking at >uncompressed HD over SDI from the J-H3.

It sure looks like some sort of ringing. Are you sure it's not the Cinewave system or your J-H3 deck adding the problem? You really need to shoot a chart and look at the cameras raw output on a studio monitor. Martin is right, start at the top and work your way down the HD chain.

Tom McDonnell
Dir/DP
New Orleans, La


Everything, including the J-H3 and the Cinewave HD-BOB (we hooked that into our SD Cinewave system) was rental.

So I'll do my best to track what I can down!

Thanks again everybody for all the help. Monday I'll check out some more stuff, do some snooping, and see what might be the culprit.

Jason Rodriguez
Post Production Artist
Virginia Beach, VA


Does this appear / disappear as you move the focus plane, and does the fringing change size when you do that ?

I'm seeing the magenta and green fringing and really, this seems like (lateral, I think) chromatic aberration to me.

No I don't why aliasing occurs along with it.

Sam Wells


Sam Wells wrote:

>Does this appear / disappear as you move the focus plane, and does >the fringing change size when you do that ?

The fringing come from using a Canon 11x4.7 zoom, which Isn't the best glass out there, not compared to an Angenieux Primo or the Zeiss Digiprimes. It's the source of the aliasing that I'm gonna have to track down the source of, but I can't do anything till Monday on that.

I'll let you guys know what I find out.

Jason Rodriguez
Post Production Artist
Virginia Beach, VA


>You are also getting color fringing on most of your edges -- magenta to >the left, green to the right. I'm assuming this is optical, but it's sure not >pretty.

--- I have to go back to this post and check out the pic again, because the above is what popped out at me right away, and apparently I mistakenly assumed it was a registration problem. (It sure wasn’t pretty.)

Jeff Kreines made a comment about that but failed to elaborate but it ***looked*** like registration to me is all I can say.

Perhaps Jeff K. would offer his wisdom and steer me straight

Jeffery Haas
freelance editor, camera operator
Dallas


Hi Jeffery,

There's definitely color fringing, if you go through some of my replies you'll see that I was only using a Canon 11x4.7 ENG zoom, but I have footage from another Canon zoom and still have the same problems.

I have some footage at work off an Angenieux HD Primo Zoom that I'll look at on Monday, but it was greenscreen, and it might not show the problems since this aliasing is on edges (or at least most apparent on edges). I'll be curious to see where this problem crops up, because the F900 was purchased by the rental company back in February (it's actually a sub-rental that another production co. gives them while they're not using it), so it's pretty new.

But thanks again everybody for all the help/feedback.

Jason Rodriguez
Post Production Artist
Virginia Beach, VA


Another Good reason for shooting the 90% Safe Area and using that as the working frame. Throw away the Ugly distorted 10% part of the frame.

If the problem can't be corrected that may be another option.

Have you tried other lenses just to check.

B. Sean Fairburn SOC
Director of Photography
Castaic Ca


Okay,

I looked at a frame from the HD-SDI out put of the camera piped into a S-Two recorder that we had on set (that worked momentarily and then had a problem with the power supply so it refused to record), and there are no artifacting problems that I can see. So it's not the camera's chips, and it's not the lens, because again, this is from an uncompressed HD-SDI output to an S-two DFR, so we're totally bypassing any HD-CAM compression.

So that leaves a problem with the recorder, or a problem with the J-H3 and/or Cinewave system. The only way to eliminate the recorder is to digitize on a different system, something I unfortunately don't have available to me.

You can see the uncompressed frame capture here (using Angenieux HD Primo) :

http://home.mindspring.com/~jrod/HDSDI.jpg

Thanks again for everyone's help,

Jason Rodriguez
Post Production Artist
Virginia Beach, VA



Copyright © CML. All rights reserved.