Texas Shootout Aliasing Problems

E

EyeReach Productions

Member
In reading the blog on HD for Indies about the Texas Shootout, I was interested in comments about the aliasing problems of the HVX. I have only had mine for a week and haven't shot much yet.

Have any of you experienced problems with aliasing?

If so, how do I avoid or minimize this aliasing affect?
 
Compiled quotes from "HD for Indies"

"I don't know which color reproduction I like best yet, I think probably the Panasonic but that's a VERY PRELIMINARY gut read, NOT based on side by side comparisons enough. But the Panasonic "just feels good," and as a post guy, I like the idea that I wouldn't have to muck around much with the colors in post, because well, the colors are already looking pretty real, as in realistic, as in what you saw with your eyeballs.

As far as codecs go, based on Adam Wilt pointing out a lot of things, I think Canon has the nicest of the HDV codecs, while DVCPRO HD is preferable in terms of fewer compression artifacts, although it suffers in terms of resolution - 1920x1080 formats are recorded as 1280x1080 (50i at 1440x1080), and 720p formats are recorded as 960x720 instead of 1280x720.

If I were doing something strictly for HDTV with a lot of fast motion and didn't have deep post budget for post and color correction, and had around $10K camera budget (assuming purchase or metric equivalent rental) I'd think about shooting HVX200 and editing natively. The good color reproduction requires less tweakage in post, the post workflow is simple and works on inexpensive equipment without requiring HD-SDI, and has low drive space and drive throughput requirements but is still a nice codec.

If, on the other hand, I was shooting something I was confident would be seen on a larger screen, and it included a lot of indoor shots or minimal fast pans or fast motion, and had an NLE that could handle 24F mode (or 1080i50 in a pinch for later conform to 24p), I'd consider the XL H1.

those are just two examples, don't rush off and make your project based on those. They ALL have various pros and cons to them that doesn't make one outstandingly better than the other at a given/similar price point.

-for home theater destined stuff, HVX200 looks awfully good "right off the truck" as they say-meaning if well shot, uncorrected, it looks awesome. MOST home theater systems would have a hard time showing the shortcomings except for some detail aliasing I THINK (need to verify more later)

-BUT for theatrical, the artifacts would DEFINITELY show if filmed out or done on proper high res truly 1080 res or 2K projector.

-XL H1 was nicely sharp, for instance, and would show better detail.

Also, specs aren't everything. While the HVX200 kept sinking in the ratings in the objective tests (Chroma DuMonde, etc.), when I saw footage I LOVED it - colors look vivid and real in CineLink-D. Now, liking the color reproduction of a camera CAN be a subjetive thing. But nobody chimed in saying they liked the HD100U, Z1U, or XL H1 color reproduction better. I likes it."

I don't know what he means by Artifacts?
They should do a film out and compare the HVX to the Canon.
But really, feature docs have been done with the DVX. Once Pana comea along with another updated camera I will add it to my collection of HVX and DVX cameras. My ten year old is now the proud op/owner of the original DVX-100. (He works really cheap!)
 
EyeReach Productions said:
In reading the blog on HD for Indies about the Texas Shootout, I was interested in comments about the aliasing problems of the HVX. I have only had mine for a week and haven't shot much yet.

Have any of you experienced problems with aliasing?

If so, how do I avoid or minimize this aliasing affect?
Click to expand...

Having been Adam's right-hand man during the Texas tests last weekend, I'll try to shed a little light on Mike's post.

The HVX suffers from aliasing in the sense that the lens resolves more than the CCD can. This means detail finer than the CCD can resolve (but still within the limits of the lens) sometimes gets distorted. The proper name for this distortion is aliasing, but just because we used the word doesn't mean it's used in the sense most do here.

When the word comes up in a forum discussion, it's usually referring to stair-stepping on diagonal lines. The HVX, as far as I saw, exhibited none of this behavior.

In other words, Adam did use the word "aliasing" in our Saturday-night quick review of footage, but didn't mean it in the sense most people here use it in. At least that is my understanding. I'll try to get some clarification from Adam sometime soon on this.
 
HVXguy said:
My ten year old is now the proud op/owner of the original DVX-100. (He works really cheap!)
Click to expand...
Good call! Get em working early and they will take our jobs one day!:)
 
nateweaver said:
The HVX suffers from aliasing in the sense that the lens resolves more than the CCD can. This means detail finer than the CCD can resolve (but still within the limits of the lens) sometimes gets distorted. The proper name for this distortion is aliasing, but just because we used the word doesn't mean it's used in the sense most do here.

When the word comes up in a forum discussion, it's usually referring to stair-stepping on diagonal lines. The HVX, as far as I saw, exhibited none of this behavior.
Click to expand...


Nate,

Thanks for that clarifacation. Stair-stepping is exactly what I was thinking, and I have not seen it on any HVX footage, so I thought something was odd about Adam's comment.

That being said, how does aliasing manifest itself? In essence, how can you tell that the lens is sharper/better than the chip? In the case of the HVX, you can't take the lens off and put it on a higher resolution body to compare it with.

A little insight would be great.

Thanks,

Rob
 
Exactly. Rodrigo. Moire.

On the HVX, when shooting a res chart, the resolution targets will show detail beyond the true resolution limit of the chips...this detail is not real, and is what most of us call moire. Adam calls it aliasing (and correctly so).

On some other cameras, there is a optical low-pass filter in front of the CCDs to filter out detail higher than the chips can resolve...this results in a much smoother and I suppose pleasing transition into high frequency detail. Cameras like the Varicam have this mastered.

(p.s. If I sound like I know what I'm talking about on this, it's because I spent 3 days by Adam's side quizzing him on stuff like this I didn't quite understand)
 
Adam and I have been trading e-mail on this point. We disagree here about the usable detail beyond the alias point. Adam contends that, as Nate says above, the detail beyond the alias point is not "real". I see it everyday in my shooting though. I am no EE but I believe the double Pixelshift is responsible for both the poor chart numbers and the wounderfull fine detial. It is true that in very high frequency situations you can see some alias, morie what ever you want to call it. So far I have seen it in exactly one shot.
That of a palmtree swaying in the very background on one wide shot that was against a white overcast sky. So very high frequency like a res chart. I think when you eliminate all the color from a scene (Res chart) you are depriving the camera of all it's pixelshifted resolution so the image aliases because the lens is resolving more than one single chip can handle. But 99% of the real image info contains color and thus resolves to the 700+ line number Barry and I have measured in static charts and test footage.
 
So it's better not to use filters if that's true, and add them in post? For instance if I wanted a scene entirely lit in red (like David Lynch) I'd light it as clean as possible then get the colorist to make it red in post with the Davinci?
 
I would say aliasing is something to be avoided. In a nutshell, it's data occurring at a frequency higher than your system is able to capture. This too-high frequency input shows up as signal at a lower frequency that your system can resolve. Thus, you capture a "false" signal. In audio its bad, in scientific data acquisition its bad, and in imaging its bad too.

Here's a site that describes it. If you don't feel like reading, at least look at the photos near the bottom.
http://www.wfu.edu/~matthews/misc/DigPhotog/alias/index.html
 
booth said:
So it's better not to use filters if that's true, and add them in post? For instance if I wanted a scene entirely lit in red (like David Lynch) I'd light it as clean as possible then get the colorist to make it red in post with the Davinci?
Click to expand...

Yes, I would do that. With all the post tools available about the only thing I might use is a grad for an overcast sky on a static shot.

On the aliasing "problem" I have been shooting with the HVX for 6 weeks now. I guess if you turn up the detail you might see some stairstepping in high detail areas. But you would see this with any video camera, especially interlaced ones. I did not see anywhere what settings these guys were using. They should really look at other HVX footage as well.
 
The aliasing that's being refered to here is not of the "stair stepping" variety.

Nice post kidtexas. That's good information.

On the HVX's defense, I have not seen any examples of this type of aliasing.
 
HVXguy said:
I did not see anywhere what settings these guys were using. They should really look at other HVX footage as well.
Click to expand...

Oh man, here we go again! From "they should really just get all the cameras together and shoot the same thing instead of looking at random web-posted footage" to "they should stop with the bad comparisons and look at what the cameras have proven to be capable of in random web-posted footage."

We're on the third cycle of this now, no?

(Nothing against you HVXGuy! The phrase in your quote just jumped out at me. Others have said the same.)

What we really need is the HVX to HVX shootout. What happened to that idea... people that see no noise and high res bring their cams, and compare to folks with the noisy low-res ones?
 
ullanta said:
What happened to that idea... people that see no noise and high res bring their cams, and compare to folks with the noisy low-res ones?
Click to expand...


This I got to see ! :)
 
ullanta said:
What we really need is the HVX to HVX shootout. What happened to that idea... people that see no noise and high res bring their cams, and compare to folks with the noisy low-res ones?
Click to expand...


Ullanta, brilliant! I would love to see a comparison between HVX's made in A, B, C, and D production months. I would give my cam as comparison. Mine's an A (Jan 06) and it had a better image then my B (feb 06).
 
ullanta said:
(Nothing against you HVXGuy! The phrase in your quote just jumped out at me. Others have said the same.)
Click to expand...

No offense taken. The HD for Indies comparision seemed frantic to say the least. I have seen no footage from it. In all of these comparisions they are always rushing and say they ran out of time. I rented the Sony and JVC cameras first. Was not impressed by either. I have looked at most of footage that Pappas posted with the Canon. If I was to make a decision based on what I see with mine own two eyes, how could anyone come to the conclusion that the Canon is worthy of spending any money on?? Sure you might get a Noisy shot or two with the HVX until you dial it in and get comfortable with it, but I would rather head that way instead of the Canon image.
 
Stevet said:
The aliasing that's being refered to here is not of the "stair stepping" variety.

Nice post kidtexas. That's good information.

On the HVX's defense, I have not seen any examples of this type of aliasing.
Click to expand...

Glad you liked it. I haven't seen it in any examples either. Then again I think the HVX, the JVC, and the Canon all rock for the money. While these test are interesting and informative, I think it has been sufficiently demonstrated that for the most part, they are all good choices, just different flavors.
 
Agreed!, they are all decent cameras.

Just think how good things are going to get in the near future.
 
Instead of people being upset with those of us who love our HVX's, go buy what works for you!
Pretty simple if you ask me.
No single tool will be right for everyone. But the HVX does it for my company and many others around here. It gives us more flexability for all our clients needs, it puts out a picture we love, and we don't have to mess with the limitations of HDV.
Again....sounds simple to me.:)
 
Totally blocking primary colors is always a bad idea in a 4:2:2, 4:2:0 or 4:1:1 system, especially in the HVX which uses the other colors for the pixel shift processing.

so if you block blue and green, using a red filter you get no pixelshift, so vertical resolution is limited to 540. Horisontal resolution is 960/2.
 
Last edited:
You must log in or register to reply here.
Back
Top