DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   Canon EOS Crop Sensor for HD (https://www.dvinfo.net/forum/canon-eos-crop-sensor-hd/)
-   -   7D Moire/aliasing ! (https://www.dvinfo.net/forum/canon-eos-crop-sensor-hd/471941-7d-moire-aliasing.html)

Daniel Browning February 10th, 2010 11:17 PM

Quote:

Originally Posted by Blake Calhoun (Post 1484454)
Here's a "real world" use of the 7D on a car lot with buildings, cars, trees, etc. - lines everywhere - and it held up real nice like me thinks.

I agree. The 720p render doesn't have nearly as much aliasing as normal 7D footage. Did you use a slow f-number by any chance (e.g. f/16)?

Brian Luce February 11th, 2010 02:00 AM

Quote:

Originally Posted by Daniel Browning (Post 1484537)
I agree. The 720p render doesn't have nearly as much aliasing as normal 7D footage. Did you use a slow f-number by any chance (e.g. f/16)?

That's a new. Didn't know 720 renders were cleaner. Supposedly native 720 (originating) is worse though.

Daniel Browning February 11th, 2010 02:20 AM

Quote:

Originally Posted by Brian Luce (Post 1484568)
That's a new. Didn't know 720 renders were cleaner.

Generally, they are. A quality downsampling algorithm will throw away both high frequency detail (resolution) as well as high frequency artifacts (e.g. aliasing). The low frequency detail and aliasing (e.g. moire) are left in place.

That said, what I actually meant by "the 720p render" was "Blake's 720p render of this specific video", to differentiate it from the lower-resolutions that Vimeo also provides. Sorry for the confusion.

Blake Calhoun February 11th, 2010 02:11 PM

Quote:

Originally Posted by Daniel Browning (Post 1484537)
Did you use a slow f-number by any chance (e.g. f/16)?

Yes I did. I had no ND so I actually shot at f/22!

And I did upload a 720p H.264 Quicktime to Vimeo, downconverted from the original 1080p file.

Daniel Browning February 11th, 2010 03:43 PM

Quote:

Originally Posted by Blake Calhoun (Post 1484767)
Yes I did. I had no ND so I actually shot at f/22!

Thanks. That explains why you didn't get any aliasing: diffraction took care of it. It was still sharp enough for 720p, though. Good work.

Manus Sweeney February 11th, 2010 03:52 PM

Quote:

Originally Posted by Daniel Browning (Post 1484537)
I agree. The 720p render doesn't have nearly as much aliasing as normal 7D footage. Did you use a slow f-number by any chance (e.g. f/16)?

i think this video looks really good too but i think it was quite an easy one for the 7d to do well.. lots of matte colours, the sky, the buildings, the cars..

the shots i've had difficulty with have been much more textured - old brick buildings, trees, landscapes, clothes textures etc

i dont think vertical lines are necessarily some awful thing to avoid but more if the frame is full of extremely detailed stuff and the camera doesnt have any chance to say 'ok over here weve just got some blue so we can use our available resolution to take care of whats in the rest of the frame'..

i guess.. or else just the colours look very matte because of the diffraction!

Brian Luce February 11th, 2010 11:59 PM

Just to summarize, a way to minimize aliasing is:

-originate in 1080
-shoot with high F stops (F16,22)
-render footage to 720p in timeline
-cross fingers

or alternately, if filming a person (as opposed to landscape/cityscape)
- originate 1080
- shoot low F stops (2.8 and below)
- render 720p

Ted Ramasola February 12th, 2010 04:53 AM

No truth to this.

I usually render to 720 my 1080 material. The aliasing found in 1080 stays in the 720.

Andy Wilkinson February 12th, 2010 06:36 AM

I agree with Ted. From my experience, unless you minimise the aliasing on the original footage (either by diffraction softening with a F stop of say F11, 16, 22 etc. or - most often a much better approach - going the other way and having a very shallow depth of field) then expect to see those horrible artifacts when problematic things are in the shot. I'd love to be proved wrong though!

Because of this I tend to use my EX3 for any overcranking work and leave the 7D in 1080p25 where it's aliasing is least problematic.

Ted Ramasola February 12th, 2010 06:46 AM

I agree that for perfections sake traditional cameras should be considered for certain scenes.

But let me share this;

In a very recent project, delivered Feb 5th, the scenario for having aliasing in scenes was very high. It was an urban real estate project.

I was aware moire and aliasing would show up in my shots due to thin lines in buildings.

I have done intensive chart tests to tell me the threshold for thin line sizes to trigger moire.

Having said that I decided to push it when I decided to shoot the wide facades of buildings with the 7D knowing it produces moire and artifacting.

I had our HD100 and HD200 with 35 adapter on standby but decided to "prove a point" that viewers don't care about us techies pixel peeping the image.

The gambit paid off cuz even though this was shown on a couple of huge screens (10.5 x 14 FEET) during the launch nobody noticed the limitations of the camera's binning method of CMOS reading.

The moire is there, so is aliasing. BUT -the viewers focused on the message, and the client(s) were very happy.

Heres the material I'm talking about.

Simply Life's Essentials http://exposureroom.com/getassetthum...4d28ba&size=sm

Daniel Karr February 14th, 2010 05:48 PM

I've been realizing more and more what you're talking about Ted. I've done 3 paid projects with my D90 in the last several months. Even though I see all kinds of technical problems with the image, the clients were thrilled with the footage. I've been going back and forth about getting a 7D, but I think if the D90 can make the clients happy, then the 7D will do just fine.

Charles Papert February 14th, 2010 06:19 PM

What it comes down to is whether the amount of aliasing is noticeable and objectionable to the client, not to us. A little glimmer here and there is not terrible--the main subject's shirt, hair or eyebrows crawling is obviously a problem. Rainbows all over a brick wall may not be noticed under certain circumstances.

The key to all this is having external monitoring of the camera. That little LCD doesn't tell enough of the story a lot of time.

Mike Calla February 14th, 2010 11:22 PM

Quote:

Originally Posted by Charles Papert (Post 1486172)
What it comes down to is whether the amount of aliasing is noticeable and objectionable to the client, not to us. A little glimmer here and there is not terrible--the main subject's shirt, hair or eyebrows crawling is obviously a problem. Rainbows all over a brick wall may not be noticed under certain circumstances.

The key to all this is having external monitoring of the camera. That little LCD doesn't tell enough of the story a lot of time.

Unfortunately, sometimes whether or not the client will accept/notice is up to client...after i spent their money:) Being up front with the client before you are hired, telling them that at certain times a different camera angle/distance/background might be needed, helps. We did a television commercial where the client/creatives were on location, and told them the moire was a problem on a certain shot and we needed to block it a little differently. They were fine with new blocking...but they didn't notice the moire in their monitor either!

If you can make beautiful images, most people won't notice at all...like me watching Ted's video...


Quote:

Originally Posted by Ted Ramasola (Post 1485121)
...The gambit paid off cuz even though this was shown on a couple of huge screens (10.5 x 14 FEET) during the launch nobody noticed the limitations of the camera's binning method of CMOS reading (...) The moire is there, so is aliasing. BUT -the viewers focused on the message, and the client(s) were very happy....[/url]

...Ted, you always make beautiful, creative films that i forgot i was watching for aliasing/moire examples... i watched a second time...and still forgot about aliasing/moire.
i didn't watch it a third time:)

Joe Lawry February 14th, 2010 11:57 PM

Can anyone tell me if shutter speed effects Aliasing?

surely a higher shutter speed would exaggerate the effect, yes/no?

Ted Ramasola February 15th, 2010 03:07 AM

@Mike, Thanks! Its important to be aware how to provoke this defect so it can be addressed ideally at shooting stage. Addressing it in post is difficult.

@Joe, Shutter speeds has no effect. Slow or fast they will be there,
if the spacing/size and angle of the design pattern meets the camera's "criteria", it will convert it to an appropriate bluish greenish mishmash pattern or moire.

It will attract attention to itself once you do a camera move. or if the part of the scene causing this moves in the frame.

This defect will lessen a bit if you turn down sharpening in your menu.

@Charles, Thats true about getting a good monitor, whats sad is some shooters might think that checking for moire using the zoom button will let them think there is non! -its a tricky electronics when zooming in on the suspected pattern gives you clear sharp and alias free image, when in fact it only aliases and moires when zoomed back and the pattern becomes small again!


All times are GMT -6. The time now is 05:02 AM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2025 The Digital Video Information Network