Rainbow-like Moire?

M

Michael Palmieri

New member
Hi,
The following link shows an example of a kind of rainbow moire pattern I shot the other day on the C300 while conducting some lens tests. I've been shooting with the camera constantly under numerous less-than-favorable conditions and this is the only instance where I've seen this problem occur, so please treat this as a very unique case-study. But this may be something to be aware of in case you decide to bring this camera into a space with a lot of stained glass, like a church.

The rainbow pattern that emerged is not the result of my ISO settings, my user settings, or anything else, but I do believe it's something that the sensor is doing. The footage was shot using Canon Raw (user settings totally disabled), utilizing multiple ISOs, multiple lenses, multiple ND settings, imported as Prores or MXF, all with the same results, which is the sort of localized rainbow moire pattern in certain areas. My hunch is that it has something to do with the quality of light coming through a stained glass window directly behind what I was shooting, but that is just a non-scientific hunch and nothing more. One other thing: NONE OF THIS SHOWED UP ON THE MONITORS AS I WAS SHOOTING IT. And I tried it on both the Canon c300 monitor and a separate SDI Monitor just to be sure it wasn't the resolution of the c300 monitor that I was up against.

The file is 35 meg, it was converted to h264 for viewing purposes but this pattern in the video is NOT an artifact of that compression. Showing this on vimeo doesn't make it as clear as what I'm seeing off the camera in its raw state so I had to post it this way instead.

Anyone have any ideas? Anyone experienced this before? It's quite a difficult if not impossible thing to remove in post, so this is why I'm asking.

Thanks, Mike

http://michaelpalmie...ainbowMoire.mov
 
Interesting. I'd say you just nailed the moire frequency with the camera. As Alister Chapman said in his review of the camera when he pushed the limits of fine detail, "Now before everyone runs off in a panic, lets put this into perspective. The F3′ aliases, the Alexa aliases as do most single chip cameras. This is certainly no worse than an F3 and is right at the resolution limits of the camera, so your not going to see it very often. It takes a very fine, high contrast pattern, in sharp focus before you’ll see this kind of thing."

I think it has a lot to do with the exceptionally high contrast on the slightly off-level horizontal lines. I'm sure you noticed already, but in the video, as you add ND and/or stop down, the moire begins to disappear.

If you have access to the same spot, I recommend to try your sharpness dialed all the way down to -10 and see if that helps alleviate the issue.

*as a point of note, there is actually no Canon RAW. I'm assuming you mean the C-Log lock. If I'm not mistaken, at C-log lock, the sharpness is set to 0. This could definitely be exacerbating the problem, especially if you're using a particularly sharp lens at optimal aperture (read Alan Roberts C300 evaluation for more info on this). Try again using C-Log gamma but sharpness dialed down.
 
Last edited:
Also, out of curiosity, what is the resolution of your SDI monitor? If less than 1080, does it have a pixel-to-pixel zoom? I try to make sure I view at 100% all sections of high contrast, fine-pattern frames with any single-sensor camera to make sure I'm not getting moire or aliasing issues. If it's coming from the sensor, it will show up in a monitor at 1:1.
 
cowpunk52 said:
*as a point of note, there is actually no Canon RAW. I'm assuming you mean the C-Log lock. If I'm not mistaken, at C-log lock, the sharpness is set to 0. This could definitely be exacerbating the problem, especially if you're using a particularly sharp lens at optimal aperture (read Alan Roberts C300 evaluation for more info on this). Try again using C-Log gamma but sharpness dialed down.
Click to expand...

Yeah, sorry, I meant C-log lock. But dialing down sharpness sounds like a plan for future avoidance.

The monitor I was using was a smallHD 8.9" so not 1 to 1 in terms of pixel mapping, which means you're right, it could have been caught on a giganto 1:1 HD monitor on set, but that's not anything any of us ever has in the field with a doc.
 
Wow, cool, thanks for this J. I'm curious if this would work for the kind of insanely sharp patterns I was seeing - I was going to note in the original post that this didn't feel like the kind of soft moire aliasing that you can sort of fix in post in an easy way -- although i've not tried your technique, it would be interesting to see if it works here. This pattern was extremely harsh in a localized area, seemed like a total roto job to me.
 
I'm 100% sure my technique will fix this.

Also I'm sure you will be startled at how effective it is !

6964875553_e7e2dfb360.jpg


BTW this looks like chromatic aberration from a lens, not moire from a sensor.
The technique I described works for CA (chromatic aberration) as well.
 
Michael Palmieri said:
Yeah, sorry, I meant C-log lock. But dialing down sharpness sounds like a plan for future avoidance.

The monitor I was using was a smallHD 8.9" so not 1 to 1 in terms of pixel mapping, which means you're right, it could have been caught on a giganto 1:1 HD monitor on set, but that's not anything any of us ever has in the field with a doc.
Click to expand...

Some monitors employ a pixel-to-pixel crop function for viewing sections of the frame at 1:1 when necessary. I know my TVLogic 5.6" does this, but I don't know if the SmallHD offers any kind of image zoom. If it does, this would a very smart feature to be aware of. I use it almost all the time on the TVLogic.

I believe the magnification function on the C300's LCD is 1:1 as well.*

*nope, it's not. It's a 2x magnification.
 
Last edited:
Ok, very cool! I'm sold :)

I would be the wrong person to ask if whats happening is chromatic aberration or not, but I did try it on a number of lenses, a leica 35mm summicron prime, an 18mm 1.4 ckbk/ekran russian prime, and a 24-105 and 17-55 canon just to make sure it wasn't the lens that was creating the problem. All showed the same moire pattern. In fact, I'd venture to say the canon lenses, due to their inherent sharpness/contrast, were the worst off in this case.

I've got another question now, esp for cowpunk52 -- why would canon have a setting on their camera designed to "disable" all of the custom settings (C-Log) for maximum post usage, but set sharpness at a level that could create this kind of problem? In other words, doesn't sharpness at 0 essentially pass thru what the sensor is picking up, doing nothing to add or subtract sharpness from what the sensor is picking up, or is this a misunderstanding of what those settings mean?

Thanks guys
 
Michael Palmieri said:
why would canon have a setting on their camera designed to "disable" all of the custom settings (C-Log) for maximum post usage, but set sharpness at a level that could create this kind of problem? In other words, doesn't sharpness at 0 essentially pass thru what the sensor is picking up, doing nothing to add or subtract sharpness from what the sensor is picking up, or is this a misunderstanding of what those settings mean?
Click to expand...

Well, the "cinema lock" setting is really meant to be used when shooting multi-cam where you want to make certain that there will be no image setting discrepancies between the cameras. And it's important to remember that sharpness at 0 is not the same as "no sharpening applied." Think of it more as "standard amount of sharpening applied." Likewise, the -10 setting doesn't mean that the image has been blurred in camera, but it is as close to native sensor performance without applied sharpening as it is possible to get. When you're recording to a compressed format, getting an image in the range of your desired final output is always a good idea, and sharpness set to 0 is probably often a nice default standard. As you said, this was the only instance in which you noticed a problem out of all your torture scenerios, and it's a pretty special case scenerio.

But just because Canon made a "cinema lock" doesn't mean that that is best for all circumstances, especially for documentary filmmakers - where shooting conditions can vary wildly from shot to shot, and being able to optimize in-cam settings for those environments will often be far more preferable to staying in a single locked CP, I would think.

I feel like this was a bit long-winded, but I hope it's a satisfactory answer.
 
I'm starting to wonder if this is a fault with the c300 sensor or a fault with your particular unit. First, I feel confident its not moire. Earlier I mentioned it looked like CA, but the more I think about it the more I'm puzzled. CA is nearly always on high contrast sharp edges like yours but all CA that I've seen is purple/green and is towards the sides of the picture where as yours is yellow/blue and in the center. Also CA doesn't shimmer and oscillate. Furthermore you stated that it was present on all lenses your tried. I'm interested to know if this is connected to in-camera sharpness settings.
 
Just happened to stumble on this thread. The Sony FS100 looks like it does exactly the same thing and i'm pretty sure it's the way that it's debayered. On very high contrast edges you can get this effect. There's an Adam Wilt review that has some images and there are a few threads about it in the FS100 forum. Apparently the F3 does not suffer the same effect because, despite the sensors being basically the same, the hardware around it is more sophisticated on the F3 and whatever this is, they've made allowances for it.

I thought the C300 was down sampled from a higher res sensor so at the same time as recognising it, i am surprised to see it here.

I get this on the FS100 even running uncompressed out, so it's not compression (as i thought originally) but something fundamentally in the way the sensor is being debayered when the pixel sites are over exposed next to pixels that aren't.

cheers
Paul
 
This look like it might be a lens issue. You know something like lens flare,no matter what lens you use,you will still get the lens flare if it's angled right to get that flare.I don't know maybe the settings will correct that?
 
Well that's a slightly 'loaded' statement.
A true statement but not necessarily related to the lens By itself, but rather, in conjunction with a given sensor.
almost all but apo lenses suffer from some chromatic aberration and based on my experience, it doesn't look like the samples offered.
Chromatic aberration in lenses is most pronounced at the periphery of the image circle diminishing to almost nothing at the center 'sweet spot'.
So if the phenomenon is consistent from corners to center then it's likely not to be a lens flaw but probably a lens + beyond DR edge contrast + a certain sensor or as one person put it, Debayering filter.
I was shocked when i jumped from film to digital at how much more prevalent this syndrome became.

It's a bit like taking two steps forward, and one step back with respect to these artifacts.

J Davis said:
I'm 100% sure my technique will fix this.

Also I'm sure you will be startled at how effective it is !

6964875553_e7e2dfb360.jpg


BTW this looks like chromatic aberration from a lens, not moire from a sensor.
The technique I described works for CA (chromatic aberration) as well.
Click to expand...
 
Zephyrnoid said:
Well that's a slightly 'loaded' statement.
A true statement but not necessarily related to the lens By itself, but rather, in conjunction with a given sensor.
almost all but apo lenses suffer from some chromatic aberration and based on my experience, it doesn't look like the samples offered.
Chromatic aberration in lenses is most pronounced at the periphery of the image circle diminishing to almost nothing at the center 'sweet spot'.
So if the phenomenon is consistent from corners to center then it's likely not to be a lens flaw but probably a lens + beyond DR edge contrast + a certain sensor or as one person put it, Debayering filter.
I was shocked when i jumped from film to digital at how much more prevalent this syndrome became.
It's a bit like taking two steps forward, and one step back with respect to these artifacts.
Click to expand...

If you read the full thread before commenting you would see I revisted at post 13 to say that in retrospect I didn't think it was CA. What do you mean by I'm making a 'loaded statement' ?
 
well, all of my settings were set to c-log off, or whatever disables the user settings altogether, as discussed above. however it does make sense that setting a lower sharpness level could reduce this problem. the biggest concern for me was i had no way of knowing this was happening while i was shooting it. it was only after downloading, viewing, and then re-shooting to see what the issue was that i came to understand the problem. i can say this though -- as the picture becomes blurrier (say, out of focus, or motion blur), the problem subsides. so, sharpness down would be a good thing.

do i think the sensor is faulty, or my particular unit is faulty? my hunch again is no. i really think this is some weird situation having to do with the stained glass fill lighting and the harsh edges that i was filming. but i'd love to see how the camera behaves in a cathedral, or a location with a lot of stained glass. that might be an interesting test. i'll try to reduplicate this, but i think it would be hard to.
 
J Davis said:
If you read the full thread before commenting you would see I revisted at post 13 to say that in retrospect I didn't think it was CA. What do you mean by I'm making a 'loaded statement' ?
Click to expand...
sorry I missed your recant. culpa mia. by 'loaded statement' I meant that there was the element of truth, but not the whole truth, prompting my analysis.
 
Switch from locked to CP 8 (Custom Picture 8). This comes standard as LOCKED but with sharpness dialled to -10. i.e. 0 sharpening. Camera at this setting is crazy sharp anyway.
 
You must log in or register to reply here.
Back
Top